Guanxi culture as a research agenda

In the framework of the contemporary Chinese institutional system, the concept of guanxi culture is commonly associated with “instrumental particularism,”Footnote ¹⁴ a term that delineates its dual aspects of instrumental and sentimental dimensions, both of which find their roots in ancient Chinese culture: Confucian renqing 人情 (meaning favour and mostly used a synonym for guanxi) embodies moral obligations and particular sentimental bonds,Footnote ¹⁵ while exclusive favouritism through sentimental guanxi arouses feelings of envy and distrust, deviating from Confucian ideals and resulting in adverse consequences for Chinese society.Footnote ¹⁶ Notably, the instrumentalism of guanxi enables it to function both as a network for individuals to access resources and as a substitute for formal institutional support in business practice. For instance, private and foreign enterprises in China rely on guanxi to engage in competition.Footnote ¹⁷ Table 1 presents the positive and negative effects of instrumental and sentimental guanxi culture, primarily derived from individual-level empirical studies.

Table 1.

The Effects of Instrumental and Sentimental Guanxi

Although this research distinguishes between instrumental and sentimental guanxi cultures for operationalization and macro-level verification, the two forms both exhibit elements of instrumentalism and particularism and coexist in a binary manner. Instrumental guanxi culture leans more towards instrumental purposes while aiming to foster particular trust. Rewards are anticipated following practices like gift giving but are not always immediate and may come after the establishment of closer sentimental bonds.Footnote ¹⁸ When such practice becomes part of local guanxi culture, the initially personalized ties transform into generalized and diffused practices, connecting even those who were once distant, such as migrants and locals. Similar to Fei Xiaotong's concept of chaxugeju 差序格局 (differential modes of association), varying distances exist between the central individual and other members.Footnote ¹⁹ Nevertheless, the particularism of guanxi is inherent from the outset, making it distinctly emblematic of Chinese culture.

This study defines regional guanxi culture as a group-level social atmosphere shaped by local guanxi-related unwritten institutions – for instance, as the local manifestation of guanxi-related customs, social institutions and psychological patterns or as the summation of guanxi-related social behaviour in a regional society. The concepts of “local” matter here because both local differences and cultural and social features mean that guanxi culture patterns display extensive variation across the regions of China. For example, people from Heilongjiang, a northern province, build conservative but deep ties and limit their business to fellow villagers and friends, both lending money and sharing clients with each other.Footnote ²⁰ On the other hand, businesspeople in economically developed southern regions, such as Guangzhou and Hong Kong, tend to establish new ties with those in other parts of the world and push the envelope of acceptable practice.Footnote ²¹

Despite the above, most empirical research treats guanxi culture as an ambiguous and homogenized background that defines social behaviour in China as a whole. As a result, little is known about the socioeconomic mechanisms and effects of regional guanxi culture. For example, quantitative researchers exploring the role of guanxi networks among the Chinese imagine a unified nationwide guanxi atmosphere,Footnote ²² or they simply use fixed regional effects in their regression models to absorb and control for cultural differences across regions.Footnote ²³ Although there is a body of qualitative literature focusing on the temporal trends of guanxi culture, such as the famous Douglas Guthrie–Mayfair Yang debate on whether guanxi has held or weakened during China's marketization, there have been few scholarly examinations of the regional differences, spatial patterns and local mechanisms of the guanxi culture as it exists across China.Footnote ²⁴

One major obstacle to guanxi culture studies is that the regional culture is elusive. The quantitative studies of guanxi in China are primarily micro-based, applying questionnaires to measure individual social networks and to characterize the interaction outcomes between ties and the labour market.Footnote ²⁵ However, generating macro variables through big data provides a novel way to measure guanxi. Researchers have used big data from Google Trends, Baidu and other search engines to construct social variables that are difficult to quantify using traditional survey methods.Footnote ²⁶ The above discussion also shows that it is feasible to use a diffused guanxi practice to proxy regional guanxi culture; this generalized purpose of establishing guanxi is the main source of search data.

Gift giving as a proxy for guanxi culture

In the realm of social interactions, the practice of gift giving expresses renqing and forms the foundation for reciprocity.Footnote ²⁷ Within China's cultural context, where reciprocity is both a cultural norm and a fundamental aspect of social interactions, gift giving has emerged as the most prevalent guanxi practice with the most likely payoff.Footnote ²⁸ Pierre Bourdieu's perspective highlights how gift giving symbolically transforms the interest exchange into “a relationship set up in due form for form's sake,”Footnote ²⁹ and in guanxi culture, this form underscores the importance of fostering enduring sentimental bonds, even when there are underlying instrumental motives.Footnote ³⁰ The tradition of gift giving in China traces its roots back to the Confucian discourse. Confucianism imposes a duty of moral and ethical reciprocity and ritualizes gift giving, which contributes to upholding social harmony.Footnote ³¹

In contrast, the gift giving practice takes on a bribery-like connotation when it is reduced to a direct and immediate exchange of material interests.Footnote ³² Thus, most literature on instrumental gift giving argues that while it fosters sentimental bonds and advances personal interests, it can also facilitate illegal transactions and hurt the market order.Footnote ³³ In the hierarchical labour market and state-led market economy, giving gifts to lindao 领导 (higher-ups – i.e. government officials or superiors within organizations) for preferential working conditions, promotion or a salary increase is the most typical instrumental guanxi practice.Footnote ³⁴ This practice is also evident in commercial activities. During the early stages of China's economic reforms, entrepreneurs used scarce premium commodities such as cigarettes to establish close collaborations with officials, balancing the need to reinforce sentimental bonds and forge enduring governmental partnerships.Footnote ³⁵ Presently, this form of gift giving is incorporated into the public relations expenses of businesses, particularly private enterprises, aiming to establish and maintain guanxi with governmental bodies and other businesses.Footnote ³⁶ This form is employed as a robustness test in this study.

In evaluating regional guanxi culture, the prevalent practice of gift giving serves as a key metric for establishing and nurturing guanxi. One approach to measuring local guanxi culture involves quantifying the instances of gift giving within a specific unit, such as a province or a county. These activities are, nevertheless, sometimes prohibited and regarded as immoral. Regular social surveys might yield biased responses owing to social desirability bias or concerns about potential repercussions, making them less reliable. Therefore, this study assesses the frequency of online searches about gift giving to higher-ups, providing insights into the prevalent instrumental guanxi culture in a given region. Additionally, search data on gift giving to zhangbei 长辈 (elders – i.e. respectable seniors with a blood bond or friendship) can be used as a proxy for sentimental guanxi culture. Elders, akin to higher-ups, serve as valuable mentors, offering guidance and support. However, interactions with elders tend to be rooted more in sentimental connections rather than instrumental motives, differentiating them from guanxi with higher-ups.

Population mobility in China

According to the Chinese Population Census, the floating population (liudong renkou 流动人口) is defined as those who have left their permanent registered residence and who are living in the survey location for more than half a year at the time of the survey; for host cities, they represent what researchers call inbound mobility.Footnote ³⁷ Since the 1950s, when China introduced the household registration system (hukou 户口), people have been considered to be permanent residents only if their household registration is formally transferred to the new location; those who do not change their registration are called migrants – or the floating population.Footnote ³⁸ It was not until the late 1970s, with the onset of market reforms, that the government's control over geographical mobility weakened and migration commenced.Footnote ³⁹ Chinese rural–urban migration garners significant scholarly attention owing to the vast inequalities in socioeconomic development between urban and rural areas in China.Footnote ⁴⁰ In comparison, urban–urban migrants frequently move across provinces, benefiting from better access to social insurance and jobs offering labour contracts.Footnote ⁴¹

The existing literature often categorizes the influencing factors of migration decisions into push versus pull, where push factors in a migrant's home region, such as low literacy and poor health, can motivate migration to a different location.Footnote ⁴² Especially in China, the surplus of rural labour is the main push factor.Footnote ⁴³ In terms of pull factors, the primary one for decades now has been the relaxation of hukou restrictions which has allowed for greater mobility across the country.Footnote ⁴⁴ The demands of the industrial sector have also served as a major pull factor. Migration tends to be motivated by a desire to maximize economic opportunities and an area's higher wages will “pull” workers from lower-paying areas.Footnote ⁴⁵ The subsequent section delves into the effects of guanxi networks in the migration process and presents the primary hypothesis.

Guanxi networks and mobility

Up until now, an extensive literature has identified that individuals base their migration decisions partially according to their guanxi networks or social capital. Chinese migrants leverage their social capital during the migration process, often associating and seeking assistance from individuals hailing from the same province and ethnic background.Footnote ⁴⁶ Social capital and guanxi networks can help migrants to secure lucrative employment and achieve a higher status in their new environment;Footnote ⁴⁷ however, migrants possess limited social capital compared to local residents. Establishing fresh social capital and building new guanxi networks incurs substantial costs,Footnote ⁴⁸ as shown by the negative effects of instrumental guanxi culture delineated in Table 1. At the macro level, the strong instrumental guanxi culture of a particular region can hinder migrants by escalating the expenses of forming the new guanxi networks needed to work and live in that region and can in turn negatively affect migrants’ fairness perception.Footnote ⁴⁹ Hence, we present the core hypothesis as follows:

In addition to the primary assumptions about instrumental guanxi culture, this research also conducts a comparative analysis of sentimental guanxi culture. According to the existing findings presented in Table 1, sentimental guanxi culture plays a more ethical and orderly role in enhancing social cohesion at the regional level.Footnote ⁵⁰ Consequently, it can be deduced that sentimental guanxi culture, exhibiting opposite characteristics to the instrumental form, attracts more migrants.

Moreover, this research undertakes a tentative mediating analysis of how instrumental guanxi culture influences mobility. Although domestic investments, foreign investments and expected income could potentially serve as mediating factors, there is limited direct evidence from existing literature to support this claim.Footnote ⁵¹ Thus, this study examines potential factors impacting income. Considering that net income equals wages minus the cost of living, income can actually act against migration.Footnote ⁵² Although this hypothesis lacks robust support from existing research, this paper presents the mediating analysis findings at the end of the results section, with detailed information available in the Supplementary Material (Appendix Part B).

Instrumental guanxi culture as the independent variable

Instrumental guanxi culture, a factor that affects the migrant population, serves as the main explanatory variable in this study. Given that the data from questionnaires cannot capture macro guanxi culture, we use local online searches related to guanxi as a proxy to quantify it. According to the China Internet Network Information Centre (CNNIC), 829 million Chinese people had used search engines by December 2021, and the user scale of Baidu reached 632 million in March 2022.Footnote ⁵³ As the most popular search engine in China, Baidu offers instant access to a gigantic ranking database of online search volumes of the most-searched for words and phrases. The data are available for different levels, including city, provincial and national levels, with different time units including daily, weekly, monthly or yearly data.Footnote ⁵⁴ Covering both PC and mobile search information, we extract the yearly Baidu indices of guanxi-related terms for all cities, prefecture-level and higher, between 2011 and 2019, to calculate a proxy for prefecture-level guanxi culture. The temporal frame of our research ends in 2019 so as to exclude the potential impact of the Covid-19 lockdown.

To choose our Baidu search terms for describing instrumental guanxi culture, we first set up a list of terms that would clearly indicate that online searchers were searching for information on how to perform instrumental guanxi practices. It is known that guanxi can be described by many other terms, including social capital, instrumental particularism, clientelism, informal ties, renqing, etc. However, formal academic and literary terminology is unlikely to reflect the language of ordinary people. Therefore, we choose to focus our search on gift giving to higher-ups (lingdao), the most common instrumental guanxi practice for building personal connections with those who have authority or access to rich resources. Specifically, we search the Baidu index for the four most-searched gift giving inquiries: gei lingdao songli song shenme 给领导送礼送什么 (what to give higher-ups as gifts); gei lingdao songli 给领导送礼 (giving gifts to higher-ups); songli gei lingdao 送礼给领导 (giving gifts to higher-ups); and lingdao songli 领导送礼 (higher-ups and gift giving).Footnote ⁵⁵

For each city, we extract the annual search volumes for each phrase and then divide them using the number of netizens (subscribers of internet services), which we accessed from The China Urban Statistical Yearbook (CUSY). Finally, to eliminate the scale bias caused by the differences in the number of keyword searches, we standardize and sum up the search volumes for all phrases. Finally, we calculate the index of instrumental guanxi culture (IGC) for each city over the calendar year as follows:

(1)$$\matrix{ {IGC_{it} = \mathop \sum \limits_{n = 1}^4 \displaystyle{{BSI_{nit}-\mu _n} \over {\sigma _n}}} \cr } $$

In Equation (1), IGC _it refers to the IGC index of city i in year t. BSI _nit is the Baidu search index of keyword n in i in year t divided by the number of subscribers of internet services in i. μ _n is the mean of n. σ _n denotes the standard deviation.

Considering the robustness of the measure, we also try three alternative methods of constructing an IGC measure. First, we use principal component analysis on the search data of the four phrases to the principal component with the maximum eigenvalue (=2.415, explaining 60.39 per cent of the variance) to predict and calculate the index of guanxi culture (IGC_1). Second, we choose six terms that are very closely related to instrumental guanxi in Mandarin (unspoken rules, contacts, circles, mianzi 面子 (face), back door and gift giving) and repeat the aforementioned procedure to obtain the IGC index (IGC_2) for each city. Third, we replace the count of internet service subscribers in BSI _nit when calculating IGC_1 and IGC_2 with the data of the three most searched terms (weather, QQ and Taobao 淘宝) on Baidu, which controls the usage of Baidu in each city, adding another variable to depict the level of internet development.Footnote ⁵⁶ Consequently, we derive IGC_3, IGC_4 and IGC_5. We fit models using all of those different variables for IGC. As we show below, the results from models using different measures of guanxi culture are very similar.

In the administrative ranking of China's cities, aside from the four municipalities directly under the central government (Beijing, Shanghai, Tianjin and Chongqing) and the special administrative regions (for example, Hong Kong and Macau), there are 333 prefecture-level administrative units, including 293 cities, 30 zhou 州 (ethnic minority autonomous prefectures such as Yanbian 延边朝鲜族自治州 in Jilin), 7 qu 区 (ethnic minority prefecture-level regions such as Ali 阿里地区 in Tibet) and 3 meng 盟 (ethnic minority prefecture-level leagues such as Alxa 阿拉善盟 in Xinjiang). Because the CUSY only includes annual statistics on netizens in 293 cities and 4 municipalities, we construct the IGC index for 297 cities altogether in mainland China, which when combined account for nearly 96 per cent of the overall population on the mainland.

Figure 1 presents the geographic distribution of the regional guanxi culture of 297 cities in China (annual average from 2011 to 2019) according to the IGC index, showing that the cities with stronger local guanxi culture are concentrated mainly in northern China. Specifically, Hohhot, Changchun, Wuwei 武威市, Ordos 鄂尔多斯市, Zhengzhou, Beijing, Daqing 大庆市, Shenyang, Jinan and Taiyuan rank in the top ten in terms of IGC. Notably, Hohhot, the capital of Inner Mongolia province, surpassed Changchun by 109.8 per cent and ranked first in the guanxi culture index. Comparatively, cities in southern China score lower in the guanxi culture. Among them, Hefei, Changsha, Nanchang, Haikou and Hangzhou recorded relatively high scores for guanxi, but they are incomparable with some of the indices in the north. Overall, our map shows that cities in northern China generally have a stronger guanxi culture than southern cities, and this trend is in line with results from extant studies.Footnote ⁵⁷

Figure 1.

Distribution of Instrumental Guanxi Culture in 297 Cities in China, 2011–2019

Figure 2 illustrates the longitudinal changes in regional guanxi culture in 297 cities from 2011 to 2019. The overall trend declines over time irrespective of municipalities, provincial capitals or ordinary cities, as evidenced by the diminishing trends of the IGC. However, the IGCs of provincial capital cities are generally higher than those of other cities. Interestingly, the temporal declining pattern differs slightly among cities. For instance, several cities in north-west China, such as Wuwei and Hohhot, saw dramatic declines in guanxi culture beginning in 2012 when the central committee of the Communist Party of China launched its famed anticorruption campaign by issuing the “eight-point frugality code.”

Figure 2.

Longitudinal Trends in 297 Cities in China.

Dependent variable and controls

The population of inbound mobility of a city (IM) is the dependent variable. The China Urban Construction Statistical Yearbook provides the annual number of migrants who have lived in the host city for more than half a year but still have not transferred their hukou to that city, which is a proper measure for the inbound population mobility. To verify the net effect of IGC on population mobility, we need to control for a set of variables that might be related to both mobility and guanxi culture in the regression model. Based on the existing literature, we control for the prefecture-level conditions of economic development, average wages, unemployment rate and human capital. We select the gross domestic product (GDP) as a proxy variable for regional economic development because it has been established as a strong predictor of inbound mobility.Footnote ⁵⁸ The unemployment rate (UEM) and wages (WG) are the primary considerations in choosing the location for inbound mobility,Footnote ⁵⁹ and they are also the potential confounders because they are both related to migration decisions and guanxi culture. The human capital (HC) in cities, with the number of college students per 10,000 in the city as a proxy variable, is an important influence on the external inflow and concentration of labour.Footnote ⁶⁰

The data on the above variables are obtained from the CUSY. All variables take natural logarithms in the analysis. Owing to missing data on relevant variables, to fit the statistical models, we construct representative panel data for 281 cities from 2011 to 2019. Note that the combined population of the 281 cities in our statistical analysis accounts for almost 91 per cent of China's overall population, suggesting that the results from our model regressions are nationally representative. The cities with missing data, marked in grey in Figure 3, have very low populations.

Figure 3.

Distribution of Inbound Mobility in 281 Cities in China, 2011–2019

Figure 3 shows the geographical distribution of inbound mobility in China. In contrast to the distribution of guanxi culture, the north–south division cannot be found here visually. Cities ranked at the top of the list include Shenzhen, Guangzhou, Dongguan 东莞市, Chongqing and Hangzhou. In fact, the average annual inbound mobility in these cities exceeds 2 million people. Consistent with the literature, in general, cities with the highest level of inbound mobility are often found to be provincial capitals or located in the Changjiang 长江 (Yangtze) and the Zhujiang 珠江 (Pearl River) deltas.Footnote ⁶¹ Finally, the selected statistics of the relevant variables among 281 cities are shown in Table 2.

Table 2.

Selected Statistics of 281 Chinese Cities, 2011–2019

Notes: Dependent and control variables have been transformed by taking natural logarithms.

Models

We apply fixed-effects models (FE) on panel data to rule out the confounding effects of the time-invariant attributes of cities. Because the data comprise a short panel, we choose the static panel model to estimate. The FE panel model of cities can then be specified as follows:

(2)$$\matrix{ {IM_{it} = \beta _0 + \beta _1IGC_{it} + \beta _2GDP_{it} + \beta _3WG_{it} + \beta _4UEM_{it} + \beta _5HC_{it} + c_i + u_{it}} \cr } $$

In Equation (2), IM _it is the dependent variable, representing the total inbound mobility logarithm) of city i in year t. IGC _it is the core explanatory variable, representing the guanxi culture level of city i in year t. c _i is the city fixed effect that does not change with time, and u _it is the random error term of the equation.

Furthermore, considering possible reverse causality, we also introduce the previous one-year values of the main explanatory variable and all control variables into the model. As is shown in Equation (3), for instance, IGC _i(t−1) is the one-year lag of IGC, representing the guanxi culture level of city i in year t-1.

(3)$$\matrix{ {IM_{it} = \beta _0 + \beta _1IGC_{i( {t-1} ) } + \beta _2GDP_{i( {t-1} ) } + \beta _3WG_{i( {t-1} ) } + \beta _4UEM_{i( {t-1} ) } + \beta _5HC_{i( {t-1} ) } + c_i + u_{it}} \cr } $$

Note that we also test a random-effects model and present the results in the Supplementary Material (Appendix Part A), and these are quite similar to the FE model results. However, Hausman's test results indicate that the FE model outperformed the random-effects model.