Riding the Trojan Horse? EU Accession and Chinese Investment in CEE Countries

ABSTRACT Chinese investment in Central and Eastern Europe (CEE) has been expanding in recent years. As the economic ties deepen, security concerns also rise with arguments decrying Chinese investment and loans as a Trojan horse aiming at buying off political influence and dividing the European Union (EU) from within. On the other hand, some recent studies suggest that the volume and threat of Chinese capital have been exaggerated by politicians from different fronts. My study contributes to the understanding of Chinese economic statecraft in CEE by examining whether it follows a ‘divide and rule’ strategy. I argue if such a strategy holds Chinese capital should increase disproportionately in CEE countries that have recently joined the EU. I apply a difference in differences (DID) design to test whether Chinese investment and trade in CEE are driven mostly by such a divisive geopolitical motivation, finding no empirical support.

However, this political and strategic concern about Chinese malign economic influence in CEE is not easy to reconcile with studies that suggest China's economic impact in the region is limited, 4 China's soft power efforts have hardly improved its image while local governments have become more suspicious of Beijing, 5 and the opportunities and threats presented by China have been exaggerated. 6While these studies have examined CEE countries' perceptions and policies toward China, we still do not know whether China is following a 'divide and rule' strategy in the region, which is particularly relevant as it underpins the geopolitical concerns and rationale discussed above and will shape an EU common foreign policy forward.
My study aims to further our understanding of the strategy of Chinese foreign policy and economic statecraft in the CEE region and more broadly toward the European countries by examining whether such a narrative of a divisive strategy from China holds up against empirical evidence.Here and throughout the paper, I define a divisive Chinese strategy narrowly as one that aims at undermining EU unity and common policy making via intensifying public economic exchanges (e.g. trade and foreign direct investment) with countries in the CEE region. 7I argue if China does pursue a 'divide and rule' strategy then we should expect Chinese investment and trade to increase disproportionately in CEE countries that are already in or have recently joined the EU since these countries can actually influence EU common policy making.Additionally, given the unanimity voting rules in several policy areas within the EU, focusing on 'buying off' a few CEE countries would be more cost-efficient if Beijing's strategy is to block undesirable initiatives from Brussels and weaken EU unity.
To empirically examine this hypothesis, I clean up and combine two project-level datasets on Chinese foreign direct investment (FDI), which identify 175 announced Chinese projects from 2000-2019.To further alleviate concerns about underreporting and also examine the implementation of Chinese investment, I also use trade data by UN Comtrade since previous research suggests bilateral trade, Chinese export in particular, can be a good indicator of its infrastructure investment. 8sing these data, I apply a difference in differences (DID) design to test whether Chinese investment and trade in CEE are driven mostly by the above divisive geopolitical motivation.Surprisingly, I find that on average there is no discernible pattern over time for investment: after a country joins the EU, it receives neither more nor less investment from China.This null finding holds across the three groups of CEE countries that joined the EU in different years since 2000.Turning to the trade patterns, I find that EU accession leads to a lower level of Chinese export in the initial years.Further differentiating the three different groups of countries, I find that this negative impact is primarily driven by countries that joined in 2007 and 2013.Taken together, the findings suggest that on average EU accession does not lead to a disproportionate increase of Chinese investment and trade toward CEE countries, casting doubt on the 'divide and rule' narrative.
This paper makes a number of contributions.First, it presents quantitative evidence in examining the 'divide and rule' narrative.The results offer additional support to some existing qualitative  One may argue that there could be other non-economic measures (e.g.public diplomacy) or clandestine ones (e.g.bribery) that the Chinese government can deploy.These possibilities would be beyond the scope of the paper.One may also argue that China could apply economic influence to dissuade candidate states from joining the EU, thereby posing a threat to the EU.However, this rationale is not divisive per se and therefore not examined.Relatedly, I do not examine whether China is pursuing an offensive (mercantile) strategy at Europe's expense.For studies that address these different perspectives, see Philip Zelikow and others, 'The rise of strategic corruption' (2020) 99 Foreign Affairs.studies questioning the empirical evidence of the narrative. 9They also align with the findings in other fields such as Chinese foreign aid, 10 the 'debt-trap diplomacy', 11 and studies highlighting the agency of local governments and business interests 12 Second, it cleans up and merges two datasets related to Chinese FDI in CEE countries.This can facilitate more empirical research of Chinese FDI both within and beyond the region.Third, the paper also holds important policy implications.The results align with the argument that Chinese economic foreign policy is more inexperienced (or fragmented) than nefarious.Framing Chinese investment as being a 'Trojan horse' or 'debt trap' exaggerates the threats and could undercut effective policy making, particularly if it dilutes the focus of strengthening one's institutions from within. 13his paper proceeds as follows.I first review the current debates concerning whether China is pursuing a divisive strategy via economic statecraft in Europe and derive the implication of such a strategy.I then present my research design to empirically test the derived hypothesis.Finally, I draw from research in other fields and discuss the academic and policy implications of the results.

Divide and Rule?
Whether China is pursuing a 'divide and rule' strategy in CEE is at the core of both recent academic and policy debates.Some researchers note that while a strong relationship between the EU and China has a compelling logic, these two 'natural partners' remain divided over 'core political values, geopolitical interests and priorities, and conceptions of world order'. 14First, China rejects many of the norms and values that the EU looks to promote.Second, China's strategic priority remains in the Asia-Pacific while the EU is focusing on its nearby region (e.g.Russia's assertiveness, fragility in Ukraine, and migration crises).Third, as an upholder of the current liberal international order, the EU remains unsure if China will embrace the order.
Further complicating the relationship is Chinese leaders' perception of the EU being a viable or attractive partner, particularly given the banking and financial crisis in the eurozone, migrant crisis, and BREXIT.Chinese policymakers appear to view the EU as 'weak, politically divided, and militarily non-influential' and Beijing therefore prefers to pursue a 'divide and rule' strategy and deal with national capitals individually. 15Under the leadership of Xi Jinping, Beijing views the EU to be only of second-order concern to China and Europe can only play a marginalized role in China's policy discussion. 16Indeed, the flagship Belt and Road Initiative (BRI) was not even mentioned at all in the 16th EU-China summit and the Meeting of Heads of Government of China and CEE Countries in November 2013 17 Against this backdrop, one could argue that pursuing a divisive strategy in CEE can better serve China's interests in the region.First, Beijing can cultivate regional elites to 'block Western initiatives that Beijing found undesirable'. 18This strategy can be assisted by increasing Chinese investment, which benefits local elites for several reasons.Existing research has shown that FDI could spur economic growth, increase tax revenues, and expand employment. 19As such, political leaders almost always value and claim credits of their competence in attracting FDI. 20Relatedly, increased inflow of FDI could generate economic excitement in the short run and enhance the support for political leaders. 21Given the Chinese government's unique role in directing its outward FDI, these regional elites would be more likely to provide Beijing with diplomatic and political support.Second, some observers argue that CEE countries are the weak spot in Europe because they still have a huge infrastructure gap compared to Western Europe and its leaders may want to circumvent political reforms by drawing from Beijing's finance. 22As such, it is also easier for China to expand its political footprint via economic foreign policy here.Viewing the CEE as 'a much safer playing field', Beijing uses its 'state-driven commercialism' to promote preferential investment and trade deals in the region. 23For instance, Benner and Weidenfeld point to the example of Hungary refusing to sign a joint letter in denouncing China's torture of detained lawyers in 2017 and claim that 'China's strategy has already yielded major political returns by weakening EU unity, especially when it comes to EU policy on international law and human rights'. 24owever, other scholars and analysts doubt the rationale or empirical evidence of a divisive strategy from Beijing.To begin with, they stress the rising importance of Europe to China, and vice versa.Despite their differences, both sides do not pose territorial threats to each other. 25There are indeed several issue areas where both sides have mutual interests in cooperation, such as policies toward Iran, climate change, and sustainable economic development.As the BRI unfolds and China's relationship with the U.S. deteriorates, the strategic role of the EU being a potential balancing partner should become more prominent for China. 26Therefore, a weaker and divided EU does not necessarily serve Beijing's interests.Indeed, some researchers contend that Beijing 'shows little interest in weakening the EU'.Rather than pursuing a divisive strategy, China's 'primary goal in CEE is to use the region as a transportation gateway and commercial platform to Western Europe's markets' which geopolitically can give the EU, Germany, and CEE more leverage in the region and counterbalance Russia. 27Relatedly, some Chinese scholars have noticed the EU's concerns and argued that CEE countries are playing a leading role in BRI because of their gateway position to Western Europe and that China 'needs to promote triple-win cooperation with' Western European countries in CEE to alleviate their concerns of a divisive China undermining European unity. 28mpirically, some recent studies suggest that the threats from China could be exaggerated.Chinese economic and political influence in the region is still limited. 29There is no evidence that China's normative influence has been internalized by CEE countries; even for Serbia, its lack of alignment with EU policies is also unlikely to be driven by Chinese pressure., 30,31 It is also not clear that China is pursuing a divisive strategy within the CEE region.For instance, during the '16 + 1' summit in 2014, China announced the establishment of a €3 billion fund to encourage Chinese investors to participate in the public-private partnerships and privatization of certain enterprises in all CEE countries. 32Some research points out that the 'divide and conquer' discourse overlooks 'the fact that the Western European countries have much deeper economic ties with China compared to CEE' and 'has a pronounced element of patronage, as it frames the CEE region as a European economic (and security) backyard through which China is trespassing'. 33aken together, while the 'Trojan horse' narrative sounds intuitive, there are also some theoretical and empirical reasons to question its validity.Directly gauging Beijing's strategic calculations may have to await future historical examinations.However, we can still gain useful insights by studying the output of the strategies, namely the exercise of Chinese economic statecraft.The CEE region provides a good opportunity to examine the question empirically given the variation of EU membership in the region.To this date, some CEE countries are still not EU members.And the current member states also joined the EU in different stages.By examining how Chinese public economic exchanges vary across these members and nonmembers, we can empirically test whether the abovementioned divisive strategy holds up.

Chinese Economic Statecraft in CEE
I argue that if China aims at undermining EU unity and common policy via pursuing a divisive policy in CEE, then Beijing should strategically allocate disproportionately more economic resources to CEE countries that are or will soon become EU members.To begin with, the Chinese government is in a unique position in deliberately shaping the foreign investment of its firms.Historically, it has been using the investment of its state own enterprises (SOEs) to carry out its economic statecraft. 34ranted, there has been more outward foreign investment by private firms in recent years.But the state can still use powerful instruments such as licensing, prohibitions, guidance, earmarked credit, and financing (which is provided by state-owned banks) to direct their investment. 35It is therefore reasonable to argue that Beijing can funnel Chinese firms' investment toward a specific country, if it wishes to do so.
Second, a divisive strategy begets funneling investment toward EU member states since these countries would actually have a say in EU policy making. 36Additionally, in several sensitive policy areas (in particular, the Common Foreign and Security Policy (CFSP)), the EU relies on unanimity voting.If China aims at buying off some EU members to veto and block policies it deems unfavorable, then it should focus on EU member states in CEE given it would be easier to translate economic influence into political leverage in this region.

30
For instance, Belgrade's support for China on issues such as Taiwan or the South China Sea disputes could easily be attributed to China's support of Serbia in refusing to acknowledge Kosovo's independence, rather than Chinese economic leverage.One can argue that China may strategically invest in certain countries before they become EU members.To be clear, even for these countries that are on good track to joining the EU, one should still expect Beijing to increase rather than reduce its investment there after EU member-ship materializes.That said, in theory we may expect that the increase in Chinese investment could begin a couple of years before formal EU accession.I also conduct robustness checks to examine this possibility.Specifically, I move the treatment year by 1 and 2 years ahead respec-tively and then rerun the models.The results are substantially similar to the main models' and are shown in the online appendix.
It is important to note two issues.First, some recent research has highlighted the fragmented nature of Chinese foreign policy. 37It is therefore worth differentiating the announced amount of Chinese investment from its actual implementation.Currently, there is no data available for how the Chinese investment projects are carried out over time.As mentioned previously, a reasonable proxy for this could be the flow of Chinese export, particularly given China's practice of using its own labor and materials in its infrastructure projects. 38Second, we shouldn't be focusing on the increase or decrease of the absolute amount of investment.This is because the amount of Chinese investment in EU members could increase simply because there are more economic opportunities there; after the accession, investment from China and other countries (such as the Western European countries) could all expand.As such, we should examine the changes in relative terms, weighing Chinese investment and export by a given country's aggregate investment and export volumes.Taken together, if the 'divide and rule' strategy holds, we should expect on average that becoming an EU member should be positively correlated with a relatively higher share of announced investment from China, after adjusting for the increase that can be attributed to the economic advantages of EU membership.And if Beijing's investment is implemented according to this strategy, then we should also expect EU membership to be positively correlated with a higher share of Chinese export.
Hypothesis 1.All else equal, becoming an EU member should be positively correlated with a higher share of announced investment from China.
Hypothesis 2. All else equal, becoming an EU member should be positively correlated with a higher share of export from China.

Research Design
For the sample, I include all CEE countries that are included within the '16 + 1' cooperation framework. 39The list of all 16 countries is shown in Table 1.The unit of analysis is country-year.I choose to study the period from 2000-2019 since 1999 is the year when China initiated the 'Go Out' policy and 2020's data could be limited and potentially biased by the impact of COVID-19.
To study the impact of EU accession, I apply a DID design.DID is particularly useful in examining the impact of a specific policy intervention.The identification strategy involves observing some units in some periods before and after they are 'treated' (e.g.EU accession).The group of units that end up being treated is called the treated group, while the rest is referred to as the comparison group.DID typically focuses on estimating the average treatment effect for the treated (ATT), which is equal to the average difference between the actual outcomes of the treated group and the outcomes they would have obtained had they not receive the treatment.The latter is a 'counterfactual' and can only be estimated.To do so, DID relies heavily on the parallel trends assumption: had the treated group not received the treatment, their path of outcomes would be the same as the comparison group's. 40sing this design, I can estimate the impact of EU accession on Chinese investment and export.extends the DID method with multiple time periods and variant treatment timing.I choose to use the inverse probability weighting for the estimation method to alleviate concerns of extrapolation.

Dependent Variables
There are two main variables examined in this paper: Chinese investment and Chinese export.As mentioned previously, the investment variable aims at capturing the announced amount of investment.I choose to also examine Chinese export data to alleviate potential concerns of underreporting.Moreover, export data can also speak to the implementation of Chinese investment (i.e.decrease in export suggests a slow down or lack of progress of Chinese infrastructure investment).
First, to code Chinese investment, I choose to combine two public datasets that record Chinese foreign direct investment (FDI) at the project level: the China Global Investment Tracker (CGIT) 42 and the Chinese Investment in Central and Eastern Europe (CICEE) dataset. 43I choose not to use the official data published by the Chinese government given the potential problems concerning data reliability, in particular Chinese multinational enterprises' practice of routing their investment via tax havens and offshore financial centers. 44Although it is not very clear how this issue plagues the official data related to CEE countries, it stands to reason that using data reported at the project level can improve the reliability. 45he CGIT data reports Chinese investment from investment and construction projects from 2005-2021.For CEE countries, there are 78 projects during this time period.One critical limitation of CGIT is that it does not report data prior to 2005, which is particularly troublesome for this study since there are many CEE countries that joined the EU in 2004.To address this limitation, I also use the CICEE dataset which reports Chinese foreign direct investment in CEE countries from 1994-2020.There are 295 projects reported in CICEE and many of them do not overlap with the projects reported in CGIT. 46To merge the two datasets, I take out projects in CICEE that are already reported in CGIT and delete the ones without investment values, which leaves me with 121 unique projects during the period.Since the quantity of investment reported in CICEE is in Euros, I use the yearly average exchange rate 47 to convert it into US dollars first and then merge the dataset with CGIT.Finally, I aggregate Chinese investment in each country by year from 2000-2019.The aggregate values over the 20 years are shown in Table 1 and we can see that Serbia, Hungary, Romania, Poland, and Slovenia are the top 5 destinations for Chinese investment.
Second, for export data I use the version of the UN Comtrade Dataset from the Atlas of Economic Complexity which applies the Bustos-Yildirim Method to clean the original data given states' inconsistent reporting practices. 48The data covers from 1962-2019 and breaks down trade in goods by different categories.I aggregate the bilateral trade data by year from 2000-2019.As Table 1 shows, Chinese export in CEE primarily goes to Poland, Czech Republic, Hungary, Slovakia, and Romania.
Finally, for both variables I weigh each country's investment and export by their respectively yearly totals to remove the time trend and weigh down higher values due to larger economic sizes and the economic advantages of EU membership.I weigh Chinese export toward a given country by the country's aggregate import volume in a year.For investment data, I use the FDI data reported by the UN 49 and weigh Chinese investment by each country's yearly flow values. 50t should be noted that FDI values could turn negative in a given year (e.g.due to disinvestment, loans from parent companies, or negative reinvested earnings).For instance, the FDI flow values in Hungary are negative in 2015 and 2016.It stands to argue that one should give Chinese investment a heavier rather than a negative weight in these years.In this regard, I add all CEE countries' yearly FDI flow by a fixed number 51 to shift all flow values to positive, which are then used to weigh Chinese investment.As such, strictly speaking, the Chinese investment variable only represents a relative share, rather than a percentage of a given country's total investment. 52

Treatment Variable
For all CEE countries, I code whether and when they joined the EU by consulting the EU enlargement timeline 53 As shown in One could be concerned about the choice of weighing investment by the yearly totals since EU membership could also mean more investment from Western European countries.In section 4.5, I provide a robustness check weighing the investment and export data by GDP instead of their yearly totals.Additionally, since the investment data are aggre-gated from announced project values, the competition from other EU countries' investment should be less of a concern.Finally, one could argue that EU membership could lead to tighter investment screening mechanisms.However, the EU wide legal framework for screening FDI was only adopted in 2019 and did not become fully operational till 2020.Indeed, some legal firms suggest that prior to 2020 FDI screening in CEE either did not exist or was of 'little practical consequence'.The distributions of these two variables are shown in the online appendix.For the rest of the paper, I would refer to these two variables as Chinese investment and export for simplicity, though one should bear in mind that these two are both in relative terms.

Control Variables
I include three control variables: political affinity, corruption level, and GDP per capita which have been used by studies of Chinese official finance and investment. 54First, I code CEE countries political affinity using the ideal point data provided by Bailey, Strezhnev, and Voeten. 55The data uses roll-call voting in the UN General Assembly from 1946-2019 to estimate each country's foreign policy preferences.I use these estimated values to calculate how closely CEE countries' foreign policy preferences align with China's by taking the absolute values of their differences.For corruption data, I use the political corruption index by the Varieties of Democracy (V-Dem) Project. 56A higher value of the index suggests the respective country is more corrupt.For GDP per capita data, I use the Word Bank's GDP data in constant 2010 USD 57 and weigh it by the Penn World Table's population data. 58ne may suggest including additional control variables. 59But given the DID design and the small sample size, including too many controls can lead to failure of estimation.Since the main purpose of including the control variables in this application is to test and see if the (conditional) parallel trend assumption can be supported, the benefits of adding more controls would be limited.Indeed, as the appendix shows, the assumption does not appear to be violated when including even fewer controls. 60he summary statistics is shown in Table 2.We can see EU member states in CEE appear to receive both relatively more Chinese investment and Chinese export.Although there is no statistical difference in terms of their political affinity with China, EU members tend to be wealthier and less corrupt.

Results
In this section, I present and discuss the results from the DID estimation for both Chinese investment and trade.I specify the model such that treatment groups (i.e.EU members) are compared to both groups of countries that are not EU members and countries that have not joined the EU yet (e.g. the same countries before they joined the EU).Given the small group sizes, estimation and inference on group-time average treatment effects may not be very reliable.As such, I follow the advice by Callaway and Sant'Anna61 and choose to just focus on the aggregated treatment effects by time    It should also be noted that the parallel assumption is, after all, an assumption.There is no guarantee that if the assumption holds before the treatment it will automatically hold after the treatment period.
periods and across different groups.Relatedly, the Indi-vidual coefficient estimates would not be particularly meaningful.Instead, I will present plots of the aggregated effects over time and across groups.As discussed below, the results for the average effects by time will also help us examine the parallel trend assumption.

Average Treatment Effects Over Time
The aggregated treatment effects by time are shown in Figure 1.The x-axis denotes the year before or after a country becomes an EU member.For instance, 0 represents the year of EU accession.1 (−1) presents 1 year after (before) a country joins the EU.For illustration purposes, I plot 5 years before and after EU accession.The y-axis denotes the average effects, i.e. the increase or decrease of Chinese investment (export) as a share of a country's yearly total.The gray lines with triangles denote the effects before a country joins the EU, while the yellow lines with circles denote the effects after EU accession.The upper panel plots the results of Chinese investment, while the lower panel shows the results of Chinese exports.
For the parallel assumption to hold, we should observe the effects before treatment to be indistinguishable from 0 (i.e. the gray lines with triangles cross the dashed line).We can see that the assumption appears to hold for both Chinese investment and export.Looking at the yellow lines with circles, the effects of EU accession appear to be statistically insignificant for investment.However, the impact on Chinese export appears to be negative in the initial 4 years.To the extent that Chinese export can be an indicator of the progress or implementation of Chinese investment, infrastructure ones in particular, the negative impact suggests that the progress of Chinese investment can be slower once a CEE country joins the EU.This appears to tally with anecdotal evidence discussed at the beginning.For instance, while China has completed part of the high-speed railway in Serbia, the progress in Hungary is much slower.

Average Treatment Effects Across Groups
Turning to aggregated treatment effects by group, I plot the results in Figure 2. The x-axis denotes the average effects while the y-axis denotes the group of countries by their respective year of EU accession.In line with the aggregated treatment effects by time, I do not find significant effects of Chinese investment (the left panel).China appears to be less likely to export to CEE countries when they become an EU member (the right panel).It appears that the negative impact of EU membership on Chinese export could be attributed to countries that join the EU in 2007 (Bulgaria and Romania) and 2013 (Croatia).This appears to align with the lack of progress of Chinese investment in these countries as well.For instance, the $15.6 billion project of the Cernavoda Nuclear Power Plant in Romania never materialized and was taken over by a U.S. consortium in 2020. 62

Visegrad Group
Existing studies suggest that among the CEE countries China appears to focus heavily on the Visegrad Group (Hungary, Poland, the Czech Republic, and Slovakia), which all joined the EU in 2004 and except for Slovakia have been reacting positively to China's divisive policy. 63,64 t is, therefore, possible that even when China does not pursue a wide 'divide and rule' strategy, it could still concentrate its effort on this smaller group of Visegrad countries.To examine this possibility, I rerun the analysis separating the group of countries that joined the EU in 2004 into two groups: the Visegrad group and the others. 65The average effects over time are similar except that the negative impact on Chinese export only becomes significant in the fourth year after EU accession.Given the similarity and that the conditional parallel trend assumption still appears to hold, I leave the plot in the appendix.Instead, I plot the group effects in Figure 3.To implement the DID estimation, I recode the second group's EU accession year as 2005 so that the did package can treat them as two separate groups.I also rerun the analysis recoding the Visegrad group's accession year as 2005 and get substantially similar results.
Importantly, the same pattern still holds: there is no statistically significant impact of EU membership on Chinese investment or export toward the Visegrad group.

Serbia
Another potential concern of the above results could be the influence of outliers.In particular, one may assume the insignificant or negative effects could be driven by China's investment in Serbia which has been surging in recent years. 66To examine this possibility, I rerun the analysis excluding Serbia from the sample.The average effects over time stay mostly the same and are shown in the appendix.I plot the group effects in Figure 4, which shows that one group's effects change.Specifically, EU accession appears to have a positive impact on Chinese investment and export in Croatia.
That said, one should exercise extra caution in interpreting this specific case.Since Croatia joined in 2013, we only have 6 years of observation for this particular group.And as researchers have shown, the DID results for small groups can be dicey especially when we add additional control variables. 67,68Moreover, for the rest of the 10 EU countries, the effects remain either negative or statistically insignificant, as in Figure 2.

Investment Diversification
It is also possible that the lack of significant effects of EU membership on Chinese investment in CEE could be driven by a higher proportion of investment from other EU members.As such, one could be concerned that weighing the investment and export data by their respective annual totals could bias against finding positive results.Therefore, I rerun the analysis weighing Chinese investment and export by each country's GDP.The same pattern for average effects over time holds.I plot the group effects in Figure 5.We can see that the main results still hold bar the Croatia case where the negative effect on Chinese export become insignificant.

Chinese Loans and Foreign Aid
Finally, one may be concerned that Chinese FDI and export do not capture the whole picture of Chinese economic foreign policy.As such, I turn to examine Chinese official finance (i.e.development projects that are funded by the Chinese government and state-owned institutions).In Table 3,    Using this dataset, I calculate the Chinese official finance commitment and completion by year for all CEE countries.In the table, I leave out CEE countries that do not receive any official finance as reported in GCDF.The aid columns present aggregates of Chinese projects that are consistent with the OECD's criteria for official development assistance and the official finance columns are aggregates for aid, debt, and other financial flows.
We can see that Chinese aid in CEE all goes to non-EU countries, 70 with the majority in Serbia.The distribution of official finance, which includes debt, appears to be less skewed and concentrated.In particular, a couple of EU countries (Bulgaria and Romania) also receive Chinese official finance, which is composed only of loans on non-concessional terms (i.e.debt).Importantly, a recent comprehensive study of Chinese official finance across the globe concludes that while '[t]he portrayal of China as a rogue donor has helped media outlets generate clicks and Western politicians stir up nationalist sentiment', there is no empirical evidence for such a narrative.Instead, the evidence suggests that 'Chinese aid is responsive to the economic needs of recipient countries' and Chinese debt is more likely to target 'poorly governed countries'. 71

Discussion
In this paper, I examine the narrative that Chinese capital in CEE can be a Trojan horse aiming at buying off political support and dividing the European Union from within.In particular, I argue if such a divisive agenda holds, then Chinese investment and trade should increase disproportionately in CEE countries that have joined the EU.I apply a DID design to test the impact of EU membership.I find no significant impact on Chinese investment and a surprisingly nega-tive impact of EU accession on Chinese export.To the extent that Chinese foreign investment is driven primarily by state-owned actors 72 and that Chinese export can be a good indicator of the implementation of its infrastructure investment, the results challenge the hypothesis that Chinese capital in CEE is driven by a 'divide and rule' strategy.
It should be noted that the Chinese government's policies, after all, are implemented by its companies.It is therefore possible that a deliberate state strategy exists but the Chinese companies intentionally choose not to follow it. 73Although many of the Chinese companies investing abroad are state-owned, they are still profit-oriented and have been reluctant to invest in the CEE region given 'it is less densely populated, cities are dispersed across a large area, there is less infrastructure, and standards of living are lower'. 74That said, it should be noted here that the data on Chinese investment are the announced amount of projects, not the actual completion.Since it is relatively easier to announce the deal of a foreign investment project and that the Chinese government 'commands uniquely powerful instruments for guiding outward FDI', 75 the lack of significant effects of EU membership on Chinese FDI cannot simply be attributed to companies shunning the central government's directives.Indeed, other alternative explanations could align better with the findings of this paper.First, it is quite likely that Beijing is focusing its economic statecraft on Western European rather than CEE countries.In particular, CEE countries do not 'provide China with primary resources or cutting-edge technology'. 76Western European countries, by comparison, have at-tracted much more Chinese investment in the first place. 77As Stroikos points out, to the extent that there was a Chinese grand strategy, its focus has been on 'the world's major powers' and in Europe the 'three key states (Britain, France, and Germany)'. 78If access to the Western European market is of any concern to Beijing, it would rather prefer to bypass the whole CEE region, 'taking its investments straight into wealthier Western Europe'. 79econd, an important reason for the null results in this paper could be that EU membership promotes democracy and strengthens transparency in candidate states. 80Garlick shows that Chinese investment in the Western Balkan countries enjoys free rein due to 'the lack of EU formalities and regulations'. 81In contrast, Chinese firms tend to face higher levels of standards and regulations in EU member states. 82For instance, the Chinese Overseas Engineering Company (COVEC)'s bid in building two sections of a major highway in Poland floundered and failed in 2011 partly due to the higher standard of EU environmental regulations and closer scrutiny of transparency and corruption. 83For another example, China has also been keen on the 350-kilometer High-Speed Railway (HSR) from Budapest to Belgrade, although it has been progressing slowly since 2017 due to the EU's concerns about transparency. 84,85urning to the policy implications, the above discussion suggests that for countries experiencing democratic backsliding, we should expect an increase of Chinese investment given shady deals are easier to materialize in these countries.For instance, at the time of writing, China's Contemporary Amperex Technology Co. Limited (CATL) announced that it would build a €7.3 billion battery plant in Hungary, the largest one ever in Europe.Notably, this happened against the backdrop of further democratic backsliding in the country.
Relatedly, we should not brush aside the agency of local governments.A recent report of Chinese influence campaigns in Hungary finds that 'China does not proactively conduct systematic malign influence operations in Hungary.Its actions are sporadic, its tools are rudimentary and toothless . . . the pro-China narrative in Hungary is shaped by the Hungarian government itself and not by the Chinese side'. 86Though not examined in my paper, the recent advancement of China-Greece relations appears to highlight a similar point.A recent study suggests that 'the bilateral relationship was largely economics-centred, propelled by the Greek crisis' and that the 'business interests associated with the Greek shipping industry have had a decisive impact on the Greek government's decision to seek closer relations with China'. 87inally, Chinese foreign investment also went through a process of learning, with the Chinese side gradually became more aware of the risks. 88After nearly a decade of offering generous loans, Chinese policy makers have started 'more rigorously evaluate new projects for financing'. 89In this regard, the hype of Chinese 'Trojan horses' appears to mirror the narrative of China's 'debt-trap diplomacy' which recent research has debunked. 90Indeed, Michael Pettis emphasized repeatedly that 'it is far more helpful to view Chinese lending to developing coun-tries as more naïve and inexperienced than nefarious'. 91aken together, the patterns of Chinese investment, export, and official finance all appear to suggest the lack of evidence concerning the narrative that Beijing is pursuing a geopolitical strategy in CEE to divide the EU from within.This is not to argue that Chinese investment holds no negative consequences 92 or that China is not seeking political leverage via economic statecraft.But a clearer understanding of Beijing's strategy holds important policy implications concerning whether an EU common policy (or any other country's foreign policy) toward China should be built more upon beating China or strengthening the institutions from within. 93
Since countries joined the EU in different years, I use the did package (version 2.1.1)in R 41 which 37 Lee Jones and Shahar Hameiri, Fractured China (Cambridge University Press 2021).38 See Zeng (n 8).See also Antoine Kernen and Katy N Lam, 'Workforce Localization among Chinese State-Owned Enterprises (SOEs) in Ghana' (2014) 23(90) Journal of Contemporary China 1053 and Jeffrey B Nugent and Jiaxuan Lu, 'China's Outward Foreign Direct Investment in the Belt and Road Initiative: What Are the Motives for Chinese Firms to Invest?' (2021) 68 China Economic Review 101,628.39 I choose to focus on the "16 + 1. 40 For more detailed explanations, see Joshua D Angrist and Jorn Steffen Pischke, Mostly Harmless Econometrics: An Empiricist's Companion (Princeton Univ press 2009).41 Brantly Callaway and Pedro HC Sant'Anna, 'Difference-in-Differences with Multiple Time Periods' (2021) 225(2) Journal of Econometrics 200.

Figure 1 .
Figure 1.Average effects time of EU accession on investment and export from China with 95% confidence interval.

Figure 2 .
Figure 2. Average effects by group of EU accession on investment and export from China with 95% confidence interval.

Figure
Figure Average effects by group of EU accession on investment and export from China with 95% confidence interval, separating the Visegrad group.
I aggregate the amount of Chinese official finance from 2000-2019 using the Aid-Data's Global Chinese Development Finance (GCDF) Dataset, Version 2.0. 69The data covers projects supported by Chinese official financial commitments from 2000-2017.Additionally, it also provides implementation details from 2000-2021.

Figure 4 .
Figure 4. Average effects by group of EU accession on investment and export from China with 95% confidence interval, excluding Serbia.

Figure 5 .
Figure 5. Average effects by group of EU accession on investment and export from China with 95% confidence interval, excluding Serbia.

Table 1 .
Chinese aggregate investment and trade with CEE in million USD (2000-2019) Tamás Matura and others, 'Chinese Investment in Central and Eastern Europe Data Set' [2021] (https://www.china-ceeinvestment.org/the-dataset, accessed May 11, 2022) Central and Eastern European Center for Asian Studies.
43 45 Jean-Marc F Blanchard, 'Chinese Outward Foreign Direct Investment (COFDI): A Primer and Assessment of the State of Cofdi Research' [2019] Handbook on the international political economy of China 76. 46 One limitation of CICEE is that many projects' financial details are not disclosed.Future work could update these projects' financial values if data become available.However, it appears that several large projects are reported in both CGIT and CICEE.For instance, Zijin Mining's investment in Serbia and 2018 and 2019 (690 and 380 million USD respectively) and Zhejiang Jinke's investment in Slovenia (1050 million).Since I aggregate the financial values on a yearly basis, the overlap of large projects should alleviate the concerns of missing data.

Table 1 ,
Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia, and Slovenia joined the EU in 2004.Bulgaria and Romania joined in 2007.And Croatia joined in 2013.The remaining (Albania, Bosnia and Herzegovina, Montenegro, North Macedonia, Serbia) are coded as non-EU countries.

Table 2 .
Summary Statistical tests performed: Wilcoxon rank-sum test; chi-square test of independence. b