3.1 Not all recommendations count

In the UPR, the recommendations made during inter-state dialogues deal with diverse human rights agendas with varying levels of severity. The contents of the recommendations are based on the UN Declaration of Human Rights and can largely be distinguished into two groups of rights: civil and political rights based on the International Covenant on Civil and Political Rights (ICCPR); and economic, social, and cultural (ESCR) rights based on the International Covenant on Economic, Social, and Cultural Rights (ICESCR). Recommendations related to political and civil rights may touch upon issues such as freedom of speech, detention, torture, and judicial rights. Meanwhile, recommendations related to economic, social, and cultural rights address issues such as education, health policies, sustainable development, employment, and poverty.

UPR statements also vary significantly in terms of severity from recommendations that merely note attention to or encourage the SUR’s human rights practices to those which urge a specific action for sensitive matters, such as abolishing the death penalty. Recognizing that state recommendations share a certain format, UPR Info, a Geneva-based NGO, has systematically categorized state recommendations into five ranks. Led by Edward McMahon and Marta Ascherio, a team from UPR Info developed Action Category, ranking recommendations on a scale of 1 (minimal action) to 5 (specific action; Mao and Sheng, Reference Mao and Sheng2017; UPR Info, n.d.). Below is the Action Category compiled by the UPR Info team, which is used to determine which recommendations are considered human rights shaming and back-patting in our operationalization.

• Rank 1: Recommendation directed at non-SUR states, calling upon the SUR to request technical assistance, or sharing information (example verbs: call on, seek, and share).

• Rank 2: Recommendation emphasizing continuity (example verbs: continue, maintain, perpetuate, persevere, persist, pursue, remain, and sustain).

• Rank 3: Recommendation to consider change (example verbs: analyse, assess, consider, envisage envision, examine, explore, reflect upon, revise, review, and study).

• Rank 4: Recommendation of an action that contains a general element (example verbs: accelerate, address, encourage, engage with, ensure, guarantee, intensify, promote, speed up, strengthen, take action, and take measures or steps towards).

• Rank 5: Recommendation of a specific action (example verbs: conduct, develop, eliminate, establish, investigate, undertake as well as legal verbs: abolish, accede, adopt, amend. Implement, enforce, and ratify).

Different ranks of action levels signify that not all recommendations contain negative tones. Statements corresponding to rank 1 or 2 action levels are rather crediting and encouraging recommendations. Conversely, recommendations including and above rank 3 involve criticism, calling for a change or, at least, considering a change to different degrees. Hence, when analysing UPR recommendations, the level of action should be considered to determine whether certain recommendations are supportive and encouraging (which we call back-patting) or censuring and offensive (which we call shaming).

Table 1 shows examples of UPR recommendations and how they are categorized according to the action level and issues. For example, Australia received three recommendations of action levels 4, 5, and 4 from North Korea during the third UPR cycle, commenting on sensitive and specific issues with forthright expressions. These recommendations are considered shaming recommendations. However, recommendations made for China encourage the current human rights practices and are thus considered as action level 2. In this study, we re-categorize the issued recommendations into a binary form, considering the different nuances in the different action levels.

Table 1. Example of UPR Recommendations from DPRK

4.1 Political affinity dictates: skin nearer than the shirt

State-to-state remarks inevitably hinge on pre-existing inter-state relations. This is because states are entities necessarily and unexceptionally in pursuit of national interests, which compels them to strive ‘not to alienate other states on which those interests depend’ (Terman and Voeten, Reference Terman and Voeten2018). To secure and further national interests, states must maintain good terms with the providers or potential providers of such goods or access to the indirect benefits of the goods. Such goods can be any material and non-material benefit, such as ideological support, political support, military alliances, economic assistance (e.g., foreign aid), technological transfer, and recognition in international society. Considering strategic relationships, therefore, states avoid ‘provoking a negative judgement’ from their partners (Terman and Voeten, Reference Terman and Voeten2018). Such consideration transpires regardless of regime type, level of democracy, or economic status of states, as every state is in a state of self-help and pursuit of interest maximization to some extent. Thus, North Korea is no exception.

The influence of political affinities in international relations, a long-standing phenomenon, has been studied from diverse perspectives and is measured in various ways. For one, scholars dedicated to democratic peace suggest how democracies are peaceful and cooperative vis-à-vis one another. From democratic peace advocates’ perspective, states are hesitant to engage in aggressive foreign policies because of their strongly intertwined political, economic, and institutional ties. Conversely, some studies suggest that like-minded states in the anti-democratic camp also cooperate. Repucci and Slipowitz (Reference Repucci and Slipowitz2022) in their Freedom House 2022 report, demonstrate that autocratic states share the common goal of preventing the spread of democracy. To this end, they supply material and ideational resources to strengthen domestic politics and shield members from external interference during political turmoil (Repucci and Slipowitz, Reference Repucci and Slipowitz2022). Thus, birds of a feather indeed do flock together in international politics.

The implications of homophily are not limited to direct state relations but also apply to state behaviour in International Organizations (IOs). Pevehouse (Reference Pevehouse2003) demonstrated that regional IOs are utilized by new democratic states to consolidate democracies, contributing to the vast literature on democratic peace. The same might apply to non-democratic states. Studies suggest that reciprocal cooperation between non-democratic states is at play in the EU, NATO the UNSC, and the UNGA (Carter and Stone, Reference Carter and Stone2015; Brookings, 2018a, 2018b; Repucci and Slipowitz, Reference Repucci and Slipowitz2022; The Diplomat, 2022). For instance, in the recent UN General Assembly Resolution voting on the suspension of Russian membership in the HRC, Russia’s friends voted against the Russian suspension or stayed mute despite the evident human rights violations committed by the Russian government against Ukraine (Wilson Center, 2022). China, Cuba, North Korea, Nicaragua, and Venezuela are among the 24 countries that voted against the resolution, and Central Asian states, such as Kazakhstan, Kyrgyzstan, Uzbekistan, Mexico, Thailand, Vietnam, India, and Sudan, are among those that cast abstentions. Choi et al. (Reference Choi, Jeong, Woo and Yim2023) empirically tested this voting result and found that sharing liberal democratic principles had the most consistent effect on the likelihood of voting for the suspension of Russian membership.

How does this insight pertain to our argument about North Korean recommendations in the UPR? As the cases indicate, political proximity mediates state interaction of any form, even in international organizations and the human rights realm. We expect that in the UPR, recommendations issued by notorious states like North Korea are also dictated by predating political affinity, which Terman and Voeten (Reference Terman and Voeten2018) identify as being at work in the UPR at a general level as politicization. Politicization occurs when international organizations are driven by relations between countries rather than by their original purpose and the issues at stake.

We expect politicization to also occur when issuing recommendations, as in the case of Terman and Voeten’s (Reference Terman and Voeten2018) model, which tests recommendation acceptance. We posit that political friends unite with each other and side with their strategic partners while political opponents go against each other. When issuing recommendations, North Korea considers its political relationship with the SUR and issues recommendations for dual purposes – rewarding friends and shaming enemies. Together, we suggest a hypothesis on political affinity that North Korea will issue more (fewer) back-patting and fewer (more) shaming recommendations to its political friends (politically distant states).

4.2 Economic sanctions as public punishment

While UN voting similarity manifests general policy-wise affinity, another proxy of inter-state relation fluctuations is economic sanctions. Taking the form of arms embargos, import/export bans, financial restrictions, and travel bans, economic sanctions have become a popular tool of coercive diplomacy (Drezner, Reference Drezner2022).

Economic coercion is particularly a relevant tool in the international human rights field, as there is virtually nowhere where human rights are severely abused or democratic rights are suspended ‘without causing a group of states to react with economic sanctions’ (Marinov, Reference Marinov2005). While the efficacy of economic coercion as a diplomatic tool is still a contentious topic, little controversy exists regarding the common target of sanctions (Afesorgbor and Mahadevan, Reference Afesorgbor and Mahadevan2016). At the same time, while studies on the the determinants and the results of economic sanctions are still inconclusive, that sanctions neverthelss are still one of the frequently used policy option is a shared knowledge underlying numerous studies (Allen, Reference Allen2008; Barry and Kleinberg, Reference Barry and Kleinberg2015; Biglaiser and Lektzian, Reference Biglaiser and Lektzian2020; Chan and Drury, Reference Chan and Drury2000; Cox and Drury, Reference Cox and Drury2006; Early and Jadoon, Reference Early and Jadoon2016; Early and Peterson, Reference Early and Peterson2022; Escribà-Folch, Reference Escribà-Folch2012; Frye, Reference Frye2019; Grossman et al., Reference Grossman, Manekin and Margalit2018; Kang et al., Reference Kang, Lee and Whang2022; McLean and Whang, Reference McLean and Whang2021; Miller, Reference Miller2014; Morgan et al., Reference Morgan, Bapat and Kobayashi2014; Park and Whang, Reference Park and Whang2018; Peksen, Reference Peksen2019; Peterson, Reference Peterson2021; Seitz and Zazzaro, Reference Seitz and Zazzaro2020; Sejersen, Reference Sejersen2021; Verdier and Woo, Reference Verdier and Woo2011; Whang, Reference Whang2010, Reference Whang2011; Whang and Paik, Reference Whang and Paik2023; Woo and Verdier, Reference Woo and Verdier2020).

By definition, sanctions aim to change or prevent the policy of a foreign state by severing economic relations with it. Thus, sanctions are most often imposed on states with conflicting interests and goals – the foes. Thus, as Drezner (Reference Drezner2000) demonstrates sanctioners are eager (reluctant) to coerce their adversaries (allies). This explains why authoritarian regimes are frequent targets of economic sanctions and threats in the real world (Morgan et al., Reference Morgan, Bapat and Kobayashi2014; Hellmeier, Reference Hellmeier2021; GSDB, 2022).

Sanctions invite resistance, in various forms, from targets. Rally effect, which is cited as a major side-effect of economic punishment, may be one domestic rationale behind this. While the assumption of rally effect is speculative, it is well established that non-allied targets resist even when they know that on average the sender is likely to continue sanctioning them despite resistance (Whang, Reference Whang2010). In addition, authoritarian governments manage to use the legitimation strategy in the face of sanctions to strengthen their rule (Hellmeir, Reference Hellmeier2021). In some instances, retaliatory sanctions are imposed from the former target to the former sanctioner (Peksen and Jeong, Reference Peksen and Jeong2022; Whang and Paik, Reference Whang and Paik2023). In this sense, economic sanctions are not merely symbolic but bring about substantial consequences including unintended and reactionary behaviours.

This lends two implications. First, North Korea is likely to react harshly to economic sanctions and sanctioners. Second, faced with such international punishment, North Korea might strategically mobilize its resources to make the political environment more favourable to rally. Given that North Korea lacks pre-existing economic relations to impose retaliatory sanctions, we contend that the UPR can serve as a useful platform to show resistance, both domestically and internationally, to sanctions against it.

5.1 Data and estimation

To empirically test the hypotheses, we obtained North Korean UPR recommendation data from three sources: Terman and Voeten (Reference Terman and Voeten2018), the UPR Info database, and UN HRC working group reports. Terman and Voeten (Reference Terman and Voeten2018) provide the first systematized dataset on UPR, compiling 56,000 recommendations from the first two UPR cycles. The third-cycle data were merged to it from UPR Info’s database to create a pooled dataset. For the ease of collecting yearly data, the data covered up to the 37^th session of the third cycle, which was held in January 2021. This was because of the limitations of the UPR Info database, whose latest dataset only covered up to the 37^th session. Although we tried obtaining data for sessions after the 37^th session, national reports or troika reports – necessary to find which state issued which recommendations during the review sessions – have not yet been compiled. Some missing values in the UPR Info data, such as the level of action in a few recent recommendations of countries, were coded by referring to the UPR Info database action category codebook. Furthermore, although Terman and Voeten (Reference Terman and Voeten2018) included Palestine, a non-member observer state, our data excluded Palestine to consider only UN member states.

Using the merged UPR dataset covering sessions 1 (April 2008) to 37 (January 2021), we reformed it into a country (North Korea)–country (SUR) dyad with a one-side fixed as North Korea. Additionally, we reorganized the data into a country-year format such that each country and North Korea’s responses in all three previous sessions are listed in the column. Thus, although the dataset covers the timespan from 2008 to 2021, it is not time-series data. Hence, the controls for the time effect were later simplified by including cycle dummies. An example of this dataset is presented in Table 2.

Table 2. Example of a dataset

The dataset comprised 537 SUR cycle dyads. Ideally, it should have 579 dyads if it fully covers the three cycles for 193 UN member countries. The fewer dyads are due to the cut timespan and to the exclusion of North Korea and Palestine as SUR.Footnote ¹ Table 3 demonstrates whether the dependent variable, the instances wherein North Korea issues recommendations, shows meaningful variation. We find that North Korea issued recommendations in 111 cases, approximately 20% of the total possible country-cycle matches, while choosing to remain silent in the rest of the cases.

Table 3. Variation in North Korea’s recommendations

Compared with countries with similar human rights reputation in the international scene, North Korea issued more frequently than Syria who issued recommendations in 101 country-cycle dyads (19%), Central African Republic 29 dyads (5%), Mali 47 dyads (9%), and Eritrea 20 dyads (4%).Footnote ² However, this number is small compared to other major rights-abusive states such as Belarus who issued recommendations in 210 country-cycle dyads (39%), or Iran who issued one or more recommendations in 250 dyads (47%), or major democratic countries like Switzerland who raised recommendation in 335 dyads (62%) and South Korea who issued recommendations in 291 dyads (54%). Although North Korea’s percentage of making recommendations is lower than sizeable authoritarian regimes like Iran or Belarus or established players of human rights like Switzerland or South Korea, it is much higher than the relatively small states with notorious human rights standing like Central African Republic, Mali, or Eritrea. From this, we can see that North Korea is far from being an inactive member state in the UPR peer review – though it may not be the most talkative one.

5.2 Dependent variables

Our empirical model includes three measures of the dependent variable, the recommendations issued by North Korea in the UPR. Noting that the recommendations are not always condemnations, we separated the positive- or neutral-toned recommendations from the negative-toned ones, recoding them into back-patting and shaming measures, respectively. As mentioned earlier, the yardstick for recording is UPR Info’s five action levels. Recommendations with action levels 1 and 2 were coded as back-patting, and the remaining from ranks 3 to 5 were coded as shaming. Although no existing literature codes UPR recommendations into binary forms of back-patting and shaming, it is commonly accepted that the action levels reasonably reflect the severity of the recommendations. Studies have used UPR Info action levels to measure the intensity of state recommendations in ordinal forms (Terman and Voeten, Reference Terman and Voeten2018; Cox, Reference Cox2020; Burger et al., Reference Burger2021; Terman and Byun, Reference Terman and Byun2022).

Previous studies neglect both theoretically and methodologically the context of recommendations made during the UPR peer review sessions by considering recommendations as some form of critical remarks on SuR’s human rights conditions. Looking into the real text of the recommendations however, one can recognize that recommendations with low action levels (1–2) are more favourable endorsements to the SuR rather than actual recommendations on its human rights practices. For example, the weakest recommendations issued by North Korea include its recommendation to Belarus to ‘continue its positive efforts aimed at raising the status of women in the society, protecting maternity and supporting the family (level 1, cycle 1)’, and Qatar to ‘expand and share good practice in implementing the “Reach Out to Asia” programme’ (level 1, cycle 3). Level 2 recommendations include, ‘Share its experiences in provision of free education to all children (Bhutan, level 2, cycle 2)’ and Continue to maintain annual economic growth of average of 8% (Myanmer, level2, cycle 2). By a simple reading of the recommendations, we can observe that low-level recommendations are not in fact criticisms but are words of approval or even praises in some cases. Thus, we argue that looking into the recommendations with the ordinal 5-scale cannot capture the different tones and quality embedded in each recommendation. We therefore suggest using severity in binary forms can distinctly show North Korea’s dual usage of UPR to send messages to friends and foes. Thus, rather than using the original 5-point scale, we re-code the scale into a binary form.

Specifically, we first create and use two binary variables: shaming_bi and backpatting_bi. They are equal to 1 if North Korea issues shaming or back-patting recommendations, respectively, and 0 if they are not in the cycle corresponding to the SUR. This measure simply shows whether North Korea voiced criticism or back-patting during the peer review session. The other two variables are count variables: shaming and back-patting. By counting the number of shaming and back-patting instances in a given cycle, we try to capture the quantitative difference between North Korea’s discriminatory recommendations towards its foes and friends. In summary, four variables are used as dependent variables, each constituting a separate model. As Models 1 and 2 utilize the binary variables, it is estimated using binary logistic regression. Models 3 and 4 with count variables are estimated using a negative binomial model. Table 4 summarizes the dependent variables, and Table 5 displays their descriptive statistics.

Table 4. Summary of dependent variables

Table 5. Descriptive statistics of dependent variables

5.3 Independent variables

We aim to test the effects of political affinity and sanctions on North Korean recommendation patterns in the UPR. Thus, our empirical model includes two independent variables. First, to gauge the effects of political affinity (Hypothesis 1) on North Korean recommendations for other UN members, we use UNGA voting similarity (Bailey et al., Reference Bailey, Strezhnev and Voeten2017). Scholars agree that UNGA voting patterns provide a good set of evidence for observing issues and alignments of states in international politics, as voting inside the UNGA is comparable and shows observable actions taken by many countries at set points in time. Thus, it has become a standard measure of policy alignment in dyadic inter-state relations (Kim and Russett, Reference Kim and Russett1996; Bailey et al., Reference Bailey, Strezhnev and Voeten2017).Footnote ³ We use the ideal point distance from the UNGA Voting Data organized by Bailey et al. (Reference Bailey, Strezhnev and Voeten2017), who constructed a state-of-the-art ideal dynamic ordinal spatial model to estimate state ideal points from 1946 to present. Ideal point distance allows us to capture the distance between each state’s position vis-a-vis a US-led liberal order; thus, it will be the state position distance between North Korea and SUR in our case. A larger value of this score represents a greater distance in foreign policy preferences between the SUR and North Korea, meaning that they possess high levels of policy disparity at an international level. Conversely, a smaller value represents political affinity and greater convergence of policy choices at the international level. Voting Distance has a mean of 1.28 and a standard deviation of 0.95. The greatest Voting Distance is found between North Korea and the United States (2.84), Israel (2.61), United Kingdom (2.13), and Syria (1.91) in 2021.Footnote ⁴

Subsequently, to test the effects of economic sanctions (Hypothesis 2), we coded a binary Ongoing Sanctions variable using the Global Sanctions Database (Felbermayr et al., Reference Felbermayr, Kirilakha, Syropoulos, Yalcin and Yotov2020; GSDB, 2022). Ongoing Sanctions measure is coded as one for an SUR that imposes individual unilateral and/or EU sanctions on North Korea. From 2008 to 2021, six countries levied unilateral sanctions on North Korea in addition to the already established UN sanctions regime. These countries are Australia, Burkina Faso, Canada, Japan, South Korea, and the US. Apart from these independent unilateral sanctions, the EU has imposed sanctions on North Korea since 2006, with all 28 EU countries participating in the EU sanctions regime against North Korea. We coded one if any of the unilateral or EU sanctions is active in the corresponding year in which the SUR participates in the UPR review session, where North Korea participates as a commenter. For example, Burkina Faso started sanctions against North Korea in 2017, which are still ongoing as of 2022. Burkina Faso participated in the UPR review session as an SUR in 2008 (cycle 1), 2013 (cycle 2), and 2018 (cycle 3). Therefore, in the dataset, the Ongoing Sanctions column for Burkina Faso would be coded 0 for cycles 1 and 2 and 1 for cycle 3. The United Kingdom, which officially exited the EU in 2020, immediately put the Democratic People’s Republic of Korea (Sanctions) (EU Exit) Regulations 2019 into force on December 31, 2020, intending to ensure that certain sanction measures relating to North Korea under the EU continue to operate effectively after Brexit (Government of UK, n.d.)

The reason for only including unilateral and EU sanctions and excluding the UNSC sanctions in the measure is simple: As the UPR covers all UN member states, each member state is a member of the UN that imposed sanctions. Since 2006 and the famous UNSC Resolution 1718, the UN has imposed strict sanctions against North Korea, which have not been lifted since then. This implies that all UN member states were considered to have imposed identical sanctions at identical points of time on North Korea, with no variation in treatment. In short, all UN member states during the 2008–2021 period would be equally coded as 1. Thus, UN sanctions were removed from Ongoing Sanctions coding because including them would make all country-cycle dyads have one in the sanction column.

This is also theoretically justifiable. Because UNSC sanctions are institutional sanctions imposed under the name of the UN, they do not readily signal a single country’s preferences or animosity, but rather signal the international community’s overall concern against the target. Thus, when a country is sanctioned by the UNSC, the targeted country is less likely to perceive it as punishment by a single or a group of enemies, but rather perceive it as a form of international punishment. However, unilateral sanction clearly exhibits who the sender is. In other words, unilateral sanctions show which state is antagonistic against North Korea, signalling evident gesture of opposition and even hostility. In this sense, unilateral sanctions are clearer signs of animosity and thus are more likely to invite hostile responses from North Korea. This is in line with our second hypothesis, which claims that imposition of sanctions prompts more shaming from the part of the target –state to the imposing state out of retributory intension. Further, not only does unilateral and institutional sanctions in addition to the UNSC sanctions present who are punishing North Korea explicitly, but also signal criticism and hostility in greater severity. This is because targeting North Korea with extra unilateral sanctions, in addition to the existing UN sanctions, sends an evident signal of animosity and clear condemnation. In this sense, it is reasonable to consider only the unilateral sanctions imposed on North Korea to test its retributory politics of human rights shamings against its sanction imposers.

As we have introduced in our second hypothesis, our two main explanatory variables – UNGA voting distance and sanctions – are closely related concepts. Greater voting distance in the UNGA resemble a distant relationship in terms of foreign policy, and such distantness correlates with a country’s likelihood to impose economic punishment against North Korea. Since having perfect multicollinearity between two variables leads to an estimation bias, we tested the correlation between the two. The correlation coefficient between Voting Distance and Ongoing Sanctions is −0.1773 and it is statistically significant at 99% level. This suggests that the countries that are further away from North Korea are indeed more likely to impose sanctions on North Korea, yet the mild nature of correlation does not cause serious methodological issues, thus we include both in our analysis.

5.4 Control variables

Our empirical models include several regressors that control for the various political and economic characteristics of countries, which may affect the likelihood and severity of recommendations received from North Korea. First, we include the Imputed Polity index (fh_ipolity2) to control for the democratic status of the SURs. Imputed Polity is a measure of the imputed average of Polity and Freedom House scores (scaled 0–10); the missing values have been ‘imputed by regressing the fhpol index on the Freedom House scores’ (The QOG Data Finder, n.d.; Hadenius and Teorell, Reference Hadenius and Teorell2007; Wahman et al., Reference Wahman, Teorell and Hadenius2013; Teorell and Wahman, Reference Teorell and Wahman2018). We use the Imputed Polity index because it provides wider coverage than other often-used measures of democracy, such as Varieties of Democracy (VDEM) or Polity IV (Coppedge et al., Reference Coppedge, Gerring, Knutsen, Lindberg, Teorell, Altman, Bernhard, Cornell, Fish, Gastaldi, Gjerløw, Glynn, Grahn, Hicken, Kinzelbach, Marquardt, McMann, Mechkova, Paxton, Pemstein, von Römer, Seim, Sigman, Skaaning, Staton, Tzelgov, Uberti, Wang, Wig and Ziblatt2022).

Second, we control for human rights conditions using the Human Rights variable retrieved from the Latent Human Rights Protection Score from Fariss et al. (Reference Fariss, Kenwick and Reuning2020). This score estimates the level of government repression based on physical repressions, such as ‘one-sided-killing’ (Fariss et al., Reference Fariss, Kenwick and Reuning2020). If North Korea participates in UPR review sessions in an objective manner and solely to improve human rights in SURs, it should shame countries with poor human rights conditions and back-patt those with good human rights records. Thus, we control for the SUR’s human rights conditions. A larger number represents a country with better human rights, whereas a smaller number indicates rights-abusive conditions.

Normalization is added as a variable to control for diplomatic relationship with North Korea. Normalization is coded as 1 if the SUR has normalized relations with North Korea in the corresponding year the SUR is going through the review process and 0 otherwise. Normalized diplomatic relations with North Korea indicate the political relevance between North Korea and the SUR, which may provoke more sensitive reactions from North Korea. Geographical distance is also controlled for in our model, reflecting the existing literature that explains geographic proximity as a crucial factor behind state behaviour in the UPR (Burger et al., Reference Burger2021). According to Burger et al. (Reference Burger2021), absolute distance and border sharing are key factors in geopolitics between states. However, UPR review sessions are held in Geneva, Switzerland, with all related parties gathered in one venue, and UPR does not entail the movement of physical goods or services. Thus, rather than controlling for metric geographic distance, considering regional membership can sufficiently control the distance aspects of state relations. To this end, we create regional dummy variables based on the UN region grouping.

Additionally, we control for economic factors, such as GDP per capita, trade volume, and foreign aid, which are expected to influence North Korea’s recommendation behaviours. We control for the level of wealth of the SUR by GDP per capita for a similar reason as controlling for the level of democracy. Wealthier countries tend to be democratic, liberal states in the Northern hemisphere or the West and have a strong correlation with our main explanatory variables. Thus, to test the authentic effect of political distance (DUNGA Voting Distance) and economic punishment (Ongoing Sanctions), we control for the wealth of the SUR. Additionally, we assume that economic dependency greatly influences dyadic relationships; the more economically dependent a state is on its counterpart, the more the state needs to consider the possibility of inadvertently ruining its relationship with that counterpart. North Korea is among the most economically isolated states in the world. Hence, its few existing economic partners in trade or donors of foreign aid are likely to be valuable suppliers of resources. This may influence North Korea’s boldness in making human rights recommendations, which could signal diplomatic aggression. Therefore, we also control for SUR and North Korea’s import and export volumes, as well as the amount of foreign aid to North Korea from the SUR. Data for GDP Per Capita, and annual Import and Export volumes are retrieved from the World Bank, and Foreign Aid data are from the IMF; we take the logarithms of all these variables. Tables 6 and 7 show the descriptive statistics of all the individual and control variables.

Table 6. Description of explanatory variables

Table 7. Summary statistics of explanatory variables

6.1 Descriptive analysis

First, we commence with a simple descriptive analysis. Table 7 presents the top 10 countries that received the highest recommendations throughout the three UPR cycles. These are a mixed group of states and do not necessarily solely include enemy or friend states. North Korea’s representative enemy states, such as the US, South Korea, and Japan, top this list, as do its friend states, such as Cuba, Vietnam, and Belarus. For instance, Cuba received the second-most recommendations during the previous three cycles. On the surface, seemingly, North Korea may issue recommendations equally and impartially to all states. However, if we explore issue areas covered by recommendations, a different story emerges. Some states, such as the U.S., South Korea, and Germany receive recommendations on demanding political and civil rights, such as detention and freedom of expression, whereas others, including Venezuela, Vietnam, and Syria receive recommendations related to economic and social rights, such as corruption, racial discrimination, education, and minority rights. This suggests the need to consider the severity of recommendations for a concrete understanding of North Korea’s politicized patterns in the UPR process.

Tables 8 and 9 illustrate the top 10 countries that received the most shaming and back-patting recommendations, respectively. The shamed list includes North Korea’s enemy states, such as the US, South Korea, Japan, Canada, Germany, and Australia. Surprisingly, it also has Cuba and Vietnam. Thus, North Korea is more inclined to shame states in a bad relationship with it; however, it also issues shaming recommendations to friends. Table 9 displays a flipped version of Table 8. The top 10 back-patted states include authoritarian states with poor human rights reputations, such as Cuba, Venezuela, China, Myanmar, and Syria. These countries are politically and socially allied with North Korea. Except for Oman, which was subjected to a recommendation requesting to ‘continue including young people in the democratic process through the commission established for this purpose’ in the second cycle, no country was back-patted for its political and civil rights progress. ‘Continue to provide comprehensive protection to citizen’s economic, social, and cultural rights’ (China, Cycle 2), ‘Share with other countries its experiences in poverty reduction and agricultural development (Vietnam, Cycle 1)’, and ‘Continue efforts for the protection of vulnerable groups affected by economic sanctions and unilateral coercive measures’ (Iran, Cycle 3) are some back-patting recommendations by North Korea. Overall, the frequency of North Korean recommendations in the UPR provides some suggestive evidence consistent with our hypotheses, with a clear distinction between the groups targeted for positive and negative recommendations.

Table 8. Who receives the most recommendations? Top 10 countries

Table 9. Who receives the most shaming? Top 10 countries

6.2. Systemic analysis

Table 10 presents the results of the statistical models. Our models robustly support our hypotheses, showing that political affinity negatively – and ongoing sanctions positively – affect the chance of shaming. Except for Model 2, political affinity affects North Korea’s issuance of recommendations, and this is statistically significant.

Table 10. Who receives the most back-patting? Top 10 countries

Models 1 and 2 use the binary dependent variables to gauge the effects of political distance and sanctions on the chances of the SUR being back-patted or shamed. In Model 1, Voting Distance is positively associated with the SUR’s likelihood of being shamed and the relationship is statistically significant at p < 0.05 level. Thus, politically distant countries are more likely to see North Korea’s shaming recommendations. Likewise, Ongoing Sanctions is positively associated with higher chances of being shamed (statistically significant at p < 0.01). Thus, the SUR with ongoing sanctions against North Korea has a higher likelihood of being targeted for human rights condemnations. The SUR’s economic relations with North Korea, including its GDP per capita, trade volume, or foreign aid, do not significantly affect mitigating or facilitating condemnations from North Korea. The human rights index does not significantly influence North Korea’s tendency to pick on certain SURs either. Thus, North Korea’s decision to shame other countries depends on its political relationships and not on the objective human rights conditions of the SUR. The closer the SUR’s foreign policy preferences are to those of North Korea, the more the SUR does not sanction North Korea, the less likely it will disparage the SUR’s human rights conditions.

Among control variables, Democracy shows statistically significant and negative associations with the dependent variables in all models at the 99% confidence level. Thus, a lower level of democracy is associated with a greater chance of not only being back-patted but also being shamed by North Korea. The more authoritarian the SUR, the more likely it is to receive both condemnation and endorsements.

Model 2 tests the effects of the independent variables on the likelihood of receiving North Korea’s positive comments–back-patting. Voting Distance shows a negative coefficient but is not statistically significant. It needs to be noted that when running this model, Ongoing Sanctions is omitted because it overlaps with countries in Western Europe and the other group. Consistent with the first model, socioeconomic factors do not significantly influence North Korea’s decision to back-pat others.

Models 3 and 4 test whether the numbers of shaming and back-patting recommendations are affected by the main independent variables. As in Models 1 and 2, the coefficients of Voting Distance and Ongoing Sanctions are positive and statistically significant, supporting Hypotheses 1 and 2, respectively. Thus, political affinity within the UNGA clearly drives the number of politically biased human rights recommendations. SURs with a closer (distant) voting distance from North Korea are likely to receive a larger number of back-patting (shaming) recommendations.

Turning to control variables, contrary to expectations, economic dependency measures (e.g., foreign aid or trade volume) do not influence the number of negative recommendations. Human rights index also does not influence North Korea’s recommendation behaviours, supporting politicized conduct in the UPR. No effect is observed for economic capacity. Meanwhile, regional groups also play a role in influencing the number of positively and negatively toned recommendations; Asia-Pacific, Eastern Europe, and Western Europe receive significantly fewer shaming recommendations in almost all three models compared to the excluded group, Africa. Statistically significant coefficients indicate that Eastern Europe and Western Europe & Other group are more likely to be subjected to fewer shaming recommendations when holding all other predictors including UNGA voting distance, sanctions, level of democracy and economic development constant. We assume this is because these regions have better human rights performance in general compared to the reference group of Africa.

How does this translate into substantive effects? Figures 1 and 2, and Table 11 demonstrate the substantive effect of political affinity on the probability of North Korea shaming and back-patting SURs. Voting Distance ranges from 0.003 (Costa Rica, 2019) to 4.66 (Israel, 2013), with 1.28 as a median value. Table 11 reports the predicted probabilities of shaming_bi and backpatting_bi when the UNGA voting distance increases by 1.0 each, with all other values fixed at the median. When Voting Distance increases from 0 to 1, 1 to 2, and 2 to 3, the probabilities of being shamed rise from 0.04 to 0.06, 0.06 to 0.10, and 0.10 to 0.17, respectively; with each increase, the probability almost doubles. When Voting Distance increases to its maximum (approximately 5), the probability of being shamed is approximately 40%. Column 3 shows that an increase in the number of shaming recommendations exhibits a trend similar to that of the binary dependent variable (shaming_bi). When the UNGA voting distance score increases by 1.0, the number of shaming recommendations doubles in every phase (e.g., from 0 to 1, 1 to 2, and so on), except for the transition from 4 to 5, which lacks statistical significance. The rise from zero to four is accompanied by an increase in the number of shaming recommendations from 0.05 to 0.53. Thus, the SUR with the most dissimilar voting pattern in UNGA from that of North Korea has 10 times more shaming recommendations from North Korea. When the distance widens further to 5, the number increases to 1.7, though the result is not statistically significant.

Figure 1. Substantive effects of voting distance on probabilities of shaming and back-patting.

Figure 2. Substantive effects of voting distance on numbers of shaming and back-patting.

Table 11. Effects of voting distance and sanctions on North Korean recommendations

Standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1.

The substantive effect of Voting Distance on the probability of being back-patted also shows a coherent pattern although it is not statistically significant. When Voting Distance increases from 0 to 2, the predicted probability of receiving back-patting recommendations decreases from 0.13 to 0.07; increasing it to 5 further reduces the probability to 0.03.

Figure 3 shows the substantive effects of sanctions on the probabilities of shaming and back-patting. While an average country with no sanctions on North Korea has a 5% probability of receiving at least one shaming recommendation from it, the probability increases to 33% if the country has sanctioned North Korea. The marginal graph on the right shows that an average country with no sanctions on North Korea has 10% probability of receiving back-patting. Meanwhile, no country has received back-patting after imposing sanctions. Thus, sanctioning countries are more likely to receive shaming recommendations from North Korea. We can assume that an SUR with unilateral sanctions on North Korea provokes a stronger feeling of animosity, consistent with our theoretical justification for excluding UNSC sanctions.

Figure 3. Ongoing sanctions and predicted probabilities (North Korea).

A clear temporal trend is noticeable, with North Korea issuing more recommendations as the cycle advances. As shown in the regression table of Fig. 1, compared to the baseline (cycle 1), North Korea shows a greater tendency to issue any recommendation rather than staying silent in cycle 2 and with greater frequency. In cycle 3, the time variable particularly affects North Korea’s shaming behaviour, significantly increasing the likelihood and number of shaming recommendations. Figure 4 illustrates the marginal effects of the cycles on the probabilities of North Korean shaming and back-patting. A clear linear relationship is observed for shaming, with North Korea actively shaming others as the cycles pass. The average country in cycle 3 has an approximately 10 times higher probability (0.28) of receiving at least one shaming recommendation compared with cycle 1(0.03). Regarding back-patting, although cycle 3 experiences a subtle decrease compared to cycle 2, it demonstrates a general increase. This may signify two aspects: first, the UPR as a mechanism may become more institutionalized; and second, North Korea may have gained momentum in using the UPR as its sphere of influence.

Figure 4. Predicted probabilities by cycle (North Korea).

Although our models take into consideration of temporal variation throughout the UPR cycles, we also factor in historical contexts of North Korean human rights by testing the effect of a watershed event in North Korean human rights: the Commission of Inquiry (COI) on Human Rights of Democratic People’s Republic of Korea. The COI established by the HRC in 2013, directed ‘to investigate, systematic, wide and grave violations of human rights’ in North Korea and to the report to the Council in 2014 (King, Reference King2024). The COI Report, which was consequently released in March 2014 shocked the international community (OHCHR, n.d.b). The Report revealed that ‘systematic, widespread, and gross human rights violations’ by the institutions and officials of the DPRK, that ‘the gravity, scale and nature of these violations reveal a state that does not have any parallel in the contemporary world’, and that ‘in many instances, the violations of human rights found by the Commission constitute crimes against humanity’ (NKDB, n.d.; Human Rights Watch, 2023). Deeply disturbed by the Report, the UN and the international community urged the situation of North Korea to be referred to the UNSC or International Criminal Court (ICC) for action.

With international community’s tense response, North Korea was clearly concerned with the impact of the COI report, which can be a critical factor influencing the upward trend in North Korea’s participation in the UPR (Yoon, Reference Yoon2024; King, Reference King2024). King (Reference King2024) reports that a noteworthy shift of North Korea’s human rights approach is shown in its’ participation in the UPR in the cycles following the COI Report release. For instance, in the final session of the first cycle that was held in 2010, North Korean officials dismissed most recommendations as inaccurate or hostile, accepting no recommendations at all, and surprisingly it did not submit the Final Outcome Report that was supposed to be debated in the following plenary session (King, Reference King2024). Immediately after the publication of the COI Report, however, North Korea submitted its 4-year-late Final Report for the first cycle, and during the second UPR, the North Koreans were witnessed to be ‘much more engaged’, ‘raising recommendations made four years earlier’ (King, Reference King2024). Also, experts report that North Koreans also became more aggressive in publicly responding on various occasions, for instance, North Korean foreign minister attended the UN high-level meetings, first time in 15 years, in September 2014 (King, Reference King2019). In the same month, North Korea released the Report of the North Korean Association for Human Rights Studies written in English, which was the detailed official response to the COI Report (King, Reference King2019).

Given that there is a reason to suspect that the COI Report pressed North Korea to respond to human rights issue and in order to explore the evolution of North Korea’s participation at the UPR, we estimate the same models including a dummy Post-COI variable to examine whether upward trend can be attributed not to political affinity and economic sanctions, but to a release of the COI Report. Post-COI takes the value of 1 if the session was held after March 2014 and 0 if not. The results are reported in Table 12. The results show a consistent support for our hypothesis that voting distance in the UNGA which represents political distantness with North Korea, and imposition of economic sanctions evidently drives the condemning recommendations by North Korea even after considering the COI Report effect (models 1 and 3). However, once we take into account the Post-COI variable, although the coefficients of Voting Distance also carry an expected direction, it loses statistical significance statistically significant at a conventional level. An interesting finding is that North Korea tends to shame and back-patt more after the COI Report regardless of other explanatory variables (models 1–3). However, COI Report did not significantly increase the number of back-patting recommendations. In model 4, none of the proposed explanatory variables had significant effect, but rather Asia-Pacific and Latin America and Caribbean regional variable seems to have absorbed the effect. Our empirics also show that after the COI Report publication, North Korea shames and back-patts more to autocratic countries. Overall, in sum, our hypotheses still hold robustly even after considering the historical context of 2014 COI Report. Yet, at the same time, our findings justify the importance of historical context in North Korean human rights issue, by showing that the landmark report of the 2014 COI did bring about substantial changes to North Koreas human rights diplomacy (Table 13).

Table 12. Substantive effects of explanatory variables on recommendation

***p <0.01, **p <0.05, *p <0.1.

Table 13. North Korea’s UPR participation after the 2014 COI report

Standard errors in parentheses/***p < 0.01, **p < 0.05, *p < 0.1.

In sum, our findings strongly support hypotheses 1 and 2 concerning political affinity and the imposition of economic sanctions, respectively. These relational ties appear to be important considerations for state delegations, as they decide whether to support the recommendations received in the UPR. Such relationship seems to persist even after considering the impact of the great turning point of 2014 COI Report.