Does Loyalty Discourage Exit? Evidence from Post-2020 Hong Kong

Abstract

In his seminal book Exit, Voice, and Loyalty, Hirschman suggests that loyal members are less likely to exit when dissatisfied with the performance of the organization. In the context of a political regime, however, we argue that loyalty may actually encourage exit because loyal members are more sensitive to the performance decline of the regime. Using an original survey conducted in Hong Kong, we show that survey respondents with a stronger local identity have greater migration intentions. We also find that the heterogeneity of perceived political changes plays a significant role as a mediator. We discuss the political implications for Hong Kong.

Keywords

Loyalty exit performance decline Hong Kong migration BNO

Introduction

In his seminal book Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States, Albert O. Hirschman suggests that members of an organization, when dissatisfied with its performance, can voice out their discontent or simply exit the organization. The choice between the “voice” and “exit” options depends on one’s loyalty to the organization. Loyal members are more prone to airing dissatisfaction and less likely to exit. Hirschman’s framework has been applied to analyze a wide range of contexts, including consumer behaviors (Duque and Lado, 2010; Kucuk, 2008), labor economics (Davis-Blake et al., 2003; Lee and Whitford, 2008), marriage (Rusbult et al., 1982), and migration decisions (Gehlbach, 2006; Pfaff and Kim, 2003).

The focus of this article is on citizens’ decision to emigrate in the context of an authoritarian regime. Our central argument is that loyalty would actually encourage exit because loyal members are more sensitive to the decline of regime performance. By loyalty, we refer to the emotional identification with an imagined community, rather than that with a regime, an administration or a leader. This emotional identification leads to more time spent on obtaining information about the regime performance and stronger reactions toward performance decline. Being informed and enraged, loyal members have strong incentive to voice out their discontent. Protesting in authoritarian regimes, however, often comes with dire consequences. When voicing out is not a viable option, some dissatisfied loyal members would opt out.

We test our argument using public opinion data from Hong Kong. In the second half of 2019, Hong Kong was engulfed in a large-scale civil unrest, which subsided in early 2020, as COVID-19 befell the city. Beijing then swiftly imposed a sweeping national security law on Hong Kong. The law provides the Hong Kong government a legal weapon to crack down on civil society. Opposition media were forced to shut down. Pro-democracy labor unions were disbanded. Many members of the opposition elite have gone into exile or to jail. The electoral rule, which had been in place for more than 20 years, was overhauled, resulting in a new system that allows for fewer popularly elected seats. Meanwhile, a massive emigration wave began in silence. The exodus is partly due to the British government’s attractive immigration program: it announced in July 2020 a special immigrant visa for Hong Kong citizens who hold the British National Overseas (BNO) passport to settle in the United Kingdom. It is estimated that 5.4 million Hong Kong citizens are eligible for this special visa. In the first 12 months since the beginning of the special immigration program, about 103,900 Hong Kong citizens applied for it.

Who are the likely outgoing Hong Kongers? Our survey data show that respondents who have a strong local identity are more willing to move abroad and more likely to discuss emigration plans with their family. Consistent with our theoretical expectation, the heterogeneity of perceived regime performance plays a significant role in driving this result: respondents with a strong local identity have decidedly negative views on recent political changes. The results of our mediation analysis show that the heterogeneity of perceived regime performance accounts for a quarter of the effect of Hong Kong identity on migration intent.

A caveat is in order. Migration is a complex, if not difficult, decision to any individual or family. In no way do we claim that political factors are the only or the most crucial reason behind Hong Kong citizens’ decision to emigrate. Yet, the findings of this study indicate that some political factors do play a fairly important role in predicting one’s migration intent.

The rest of this article is organized into six parts. In section “Loyalty and exit in authoritarian regimes,” we will present our central argument. We provide background information on Hong Kong’s recent political development and emigration wave in section “The Hong Kong case: local identity, national security, and emigration.” We state our hypotheses in section “Hypotheses,” and discuss the data and operationalization in section “Data and operationalization.” Section “Results” shows the results of our statistical analysis. We discuss the implications of our argument in the last section.

Loyalty and exit in authoritarian regimes

Simple and elegant, Hirschman’s Exit-Voice-Loyalty (EVL) framework has been applied by economists, sociologists, and political scientists to analyze many types of human interactions. Its limitation, however, also lies in its simplicity. Because the EVL dynamics can vary significantly across contexts, Hirschman’s framework is unable to produce universal predictions for different scenarios. For instance, the cost of voicing dissatisfaction with a consumer product may be 30 minutes of typing a complaint email. In many authoritarian regimes, however, voices of dissent come with dire, if not fatal, consequences, a price so high that few citizens are willing to pay.¹

Equally complicated is the idea of exit. Economists have long studied exit as a market mechanism. For example, firms that lose money will exit a market. Consumers will exit a product with which they are dissatisfied. In an efficient market, switching a product brand involves negligible cost on the part of the consumers. But the story is quite different in an authoritarian context. For one thing, authoritarian governments have both legal and administrative tools to prevent their subjects from exiting the country. The Berlin Wall, which was built by East Germany to restrict citizens’ movement, was a case in point. Of course, physical hurdles aside, migration also entails significant psychological cost. This is so partly because migration often implies giving up one’s original nationality, which is an integral element in the formation of one’s social identity (Berry, 1997; Colic et al., 2003). To some, the cost of reconstructing one’s social identity as a result of migration is prohibitively high.

The most controversial element of Hirschman’s framework is the concept of loyalty (Barry, 1974; Birch, 1975; Dowding et al., 2000; Gehlbach, 2006). In economics applications, the concept of loyalty is quite straightforward. Brand loyalty, for example, refers to the psychological attachment to a brand, which is associated with behavioral implications such as repeated consumption and reluctance to brand switching. In politics, however, loyalty is a multifaceted concept (Birch, 1975). A revolutionary soldier may be loyal to the revolutionary cause, but not to the revolutionary leader, or vice versa. In the context of contemporary China, many Chinese join the Chinese Communist Party (CCP) for a better career, not necessarily because they are loyal to the Party (Dickson, 2016).

The discussion above suggests that the concept and operationalization of exit, voice, and loyalty are context-specific. Conceivably, the interactions between these three concepts also likely differ from case to case. Most notably, when Hirschman conceptualizes the relationship between loyalty and exit, he drew insights from consumer behaviors in free market. Loyal customers, by definition, are less inclined to brand switching. Extending this observation, Hirschman (1970) suggests that “[c]learly the presence of loyalty makes exit less likely” (p. 77).

Intuitive as it is, Hirschman’s observation, which can be aptly applied to explain market behaviors, unlikely carries over to the context of an authoritarian regime. In particular, we argue that loyalty, when defined as a psychological attachment to a country or an imagined community that people share emotional bonding, as opposed to loyalty to an individual leader or to the ruling elite, actually encourages exit in authoritarian regimes. To see this, we highlight two important differences between loyal and less loyal members regarding the perception of performance decline.

First, compared with those who are less committed, loyal members of a group² tend to be more sensitive to the change of the group, including its performance. From a sociological perspective, groups are sustained by rituals and traditions (Sosis and Ruffle, 2001), which reinforce ingroup identity and emotional bonding (Stein et al., 2021). Those who perform rituals religiously can establish a good reputation among the peers (Power, 2017). Loyal members, given their commitment to the group, should be more familiar with and attuned to the associated rituals and traditions, which also implies that they are more aware of the deviation from the modus operandi. These individuals should be the first to detect any performance decline pertaining to the group. For instance, the fans of an online game are able to articulate all the subtle changes of the game’s new update even before its official release because they would look for such information on some fandom video channels.

From a psychological perspective, loyalty is a form of psychological attachment (Bennett & Durkin, 2000; Meyer and Herscovitch, 2001). Psychologists have long shown that attachment is closely connected with the feeling of security, because the attachment system is evolved to deal with external threats (Bowlby, 1973, 1980). Individuals who score high on attachment anxiety tend to play the role of guardians for the group, as they constantly on the lookout for the sign of danger and threat against group survival (Ein-Dor and Hirschberger, 2016).

The second important difference between loyal and less loyal members lies in their exposure to group-related information. Studies have shown that people tend to have an ingroup bias with respect to news consumption (Hoffner and Rehkoff, 2011). For instance, Knobloch-Westerwick and Hastall (2010) show that German senior citizens preferred consuming positive news about older people and negative news about younger generations. Similarly, Appiah et al. (2013) find that African Americans spent more time-consuming positive news about African Americans and negative news about Whites. In a further study, Amira et al. (2021) provide experimental evidence showing that when given a choice between positive ingroup news and negative outgroup news, respondents would prefer the former to the latter. These studies demonstrate the importance of social identity in predicting one’s information consumption. Suffice it to say, people who have a strong social identity tend to consume more information related to that identity. If loyalty implies a strong group identity, we have reason to believe that loyal members tend to consume more group-related information than less loyal members.

The aforementioned differences between loyal and less loyal members contribute to the heterogeneous perceptions of performance decline. Eager to look for information about the group and knowledgeable about the group affairs, loyal members are more sensitive to the decline of group performance. Translating this into a political context, citizens who are psychologically attached to a political identity, be it national or partisan, tend to be more perceptive and capable of discerning changing conditions of their affiliated political group.

Under normal circumstances, loyal members are poised to voice out when they detect performance decline, as predicted by Hirschman. In many authoritarian regimes, however, the cost of voicing out is prohibitively high. While it is possible that some loyal members are so insensitive to personal risk that they would express their voice of dissatisfaction regardless, most loyal members are forced to keep quiet. For the latter, they endure two kinds of torment. The first is that they witness the deterioration of their group, while the second is that they are unable to share their discontent. In some repressive regimes, citizens even need to pretend to be undisturbed by the performance decline, in order to avoid adverse political consequences. All these suggest that in autocracies, citizens with a strong political identity likely suffer greater emotional and psychological distress than those with a weaker identity. It is such emotional and psychological distress that motivates these “loyal” citizens to choose to leave the country when this option is available. By contrast, citizens who are less “loyal” are unlikely to opt for exit because they may not even be aware of the performance decline.

The Hong Kong case: local identity, national security, and emigration

Hong Kong used to be called a “refugee society” (S-L. Ku, 2004; Lau & Kuan, 1988; Wong, 1988), because when Hong Kong was a British colony, it provided a safe haven for many Chinese who fled the country during various waves of political upheavals. As Hong Kong’s economy took off in the 1960s and 1970s, these Chinese immigrants chose to stay. The second-generation Hong Kongers saw themselves as the occupants of the city, rather than some refugees from mainland China (Ku, 2004; Lui & Chiu, 1999). Local identity gradually emerged, but not strong enough to engender the demand for self-determination. Most Hong Kong people at that time still saw themselves as Chinese citizens living in a British colony.

In the 1980s, Beijing and London negotiated Hong Kong’s sovereignty transfer. The majority of the Hong Kong citizens were averse to the idea, but they had no say in the process. Many decided to leave the former British colony, for fear of falling prey to a regime from which they had once escaped (Wong, 2012). To restore the confidence crisis, Beijing promised to preserve Hong Kong’s capitalism and civil liberties for at least 50 years. It even permitted limited political freedoms unseen in other parts of China. In particular, Hong Kong people were allowed to elect half of the members of the local legislature and to preserve its common-law judicial system that was reputed to be independent from executive interference. In addition, Beijing pledged to allow Hong Kong people to elect, in due course, the entire legislature and the Chief Executive by universal suffrage, a promise written into the mini-constitution of the city.

In 1997, the Hong Kong Special Administrative Region (HKSAR) of the People’s Republic of China was born. The special status conferred upon Hong Kong people did not make them closer to mainland China. Rather, it reinforced their distinct identity as Hong Kongers. Hong Kong people continued to enjoy a high degree of freedom of speech, freedom of assembly, and freedom of the press that existed nowhere in mainland China. That said, for more than two decades after the sovereignty transfer, most Hong Kong people saw themselves as ethnic-Chinese. As shown in the longitudinal data of the Public Opinion Program of the University of Hong Kong, most Hong Kong respondents would choose an ethnic identification with a reference to China, including “Chinese in Hong Kong,” “Hong Kong Chinese,” or simply “Chinese,”³ although the number of respondents who choose “Hong Kongers” have gradually increased over time (HKUPOP, 2019).

Pro-democracy citizens took advantage of the civil liberties to articulate their demands for democratization. Beijing, however, was reluctant to fulfill its promise for universal suffrage, resulting in growing state–society tensions in Hong Kong. As more Hong Kong people came to believe that Beijing’s political interest was at odds with theirs, an anti-China sentiment simmered in the city (Ma, 2015). Localism gradually emerged as a nascent yet powerful force in party politics and social movement (Kaeding, 2017; Veg, 2017; Wong and Wan, 2018). Some localist groups openly advocated self-determination and independence.

Beijing also seemed to lose the patience of dealing with Hong Kong citizens’ persistent demand for democracy. On 31 August 2014, the Standing Committee of the National People’s Congress announced a new set of stringent rules (also known as the August 31 Decision) regulating the future elections of Hong Kong’s Chief Executive, dashing the hope of many pro-democracy citizens. In protest against the August 31 Decision, many Hong Kong citizens took to the streets, culminating into the Umbrella Movement.

Although the Umbrella Movement failed to extract any concession from Beijing, localism continued to gain traction in the city. Some localists even got elected to Hong Kong’s legislature in 2016. In response, the Hong Kong authorities endeavored to keep localism at bay, while curtaining the influences of the larger democracy movement. For instance, the Electoral Affairs Commission banned some localists from standing elections. Social activists were arrested for their involvement in the Umbrella Movement and other protests.

The most controversial move came in the early 2019, when the Hong Kong government proposed an amendment to the extradition law, allowing extraditions to mainland China. This amendment, widely seen as transgression of Hong Kong’s time-honored judicial independence and civil liberties, ignited a massive wave of street-level protests known as the Anti-Extradition Law Amendment Bill (Anti-ELAB) Movement. “Hong Konger” as a social identity became a key movement frame, because many believe that civil liberties and judicial independence were two defining characteristics of the city (Lee et al., 2019). Losing them, Hong Kong is no different from other Chinese cities, which also implies the end of the Hong Kong identity.

Unlike the tactics employed to contain previous mass protests,⁴ the Hong Kong government this time chose to sustain a heavy-handed approach. The police’s liberal use of teargas, rubber bullets, and batons, together with the government’s refusal to conduct an independent investigation of potential police brutality, contributed to violence escalation and collective trauma. Some observers contend that this collective trauma connected all walks of life in Hong Kong, giving birth to an imagined community of shared sufferings (Leung, 2019).

Deeply dissatisfied with the government’s performance, many Hong Kongers decided to voice out their discontent by voting for the opposition in a local election that took place in November 2019. The unprecedented high voter turnout (71.2%) brought the pro-democracy opposition a landslide victory (Wong and Chan, 2021). Street-level protests continued for sometime and subsided in early 2020, when the COVID-19 pandemic hit the city.

The mobilization power of the civil society, the degree of the protest violence, extraordinary international attention, and perhaps, more importantly, the enormous public support of the Anti-ELAB movement shocked Beijing. Once the street-level protests subsided, Beijing took steps to entrench its control of Hong Kong’s restive population. In June 2020, the Chinese government decided to bypass the local legislature to impose the “Law of the People’s Republic of China on Safeguarding National Security in the Hong Kong Special Administrative Region” (the Hong Kong NSL). The Hong Kong NSL, which criminalizes secession, subversion, terrorism, and collusion with foreign organizations, has provided the Hong Kong government a legal weapon to suppress the opposition and the democracy movement at large. Opposition media and labor unions were forced to shut down. A large number of pro-democracy activists and student leaders were arrested on the charges pertaining to the Hong Kong NSL.

These high-profile arrests alarmed the international community. In particular, the British government criticized the Hong Kong NSL for infringing the political freedom and civil liberties that are protected in the Sino-British Joint Declaration. Australia, Canada, and the United States also expressed similar concerns. In response to the changing conditions in Hong Kong, the British government promulgated a special immigrant visa for Hong Kong people who held the British National (Overseas) passport (commonly known as the BNO passport). With this special visa, BNO holders can settle in the United Kingdom together with their immediate family members. It is estimated that 5.4 million of Hong Kong people are potential beneficiaries of this special visa. In July and November 2020, respectively, Australia and Canada also announced a special immigration pathway for younger Hong Kong citizens (Table 1 shows all the special immigration programs for Hong Kong citizens). In other words, young people who were born after 1997 (hence not eligible for the BNO passport) and who have no BNO-holding family members can also have a chance to move abroad. Suffice it to say, by 2021, the majority of Hong Kong citizens have various means to exit from the city, if they so desire.

Table 1.

Special immigration programs bespoke for Hong Kong citizens.

	United Kingdom	Australia	Canada
Announcement date	20 July 2020	9 July 2020	12 November 2020
Implementation date	31 January 2021	5 March 2022	8 February 2021
Eligibility	BNO citizens and family dependents (spouse or partner and children aged under 18/born after the Handover)	BNO or HKSAR passport holders (5-year visa) who are students completing tertiary education in Australia or skilled workers with a temporary skills shortage visa	BNO or HKSAR passport holders holding temporary resident status who are degree or post-secondary diploma holders (from a Canadian tertiary learning institution or an equivalent abroad on the approved list) Degree must be obtained 3 years (for graduates from a Canadian institution to apply for a postgraduate degree) or 5 years (for applicants with 1-year full-time work experience in Canada within the 3 years before) before applying Spouses and dependent children can also apply for a work or study visa
Restrictions	Ability to accommodate and support themselves in the United Kingdom for 6 months Good knowledge of the English language when applying for settled status	Health, character, and security requirements for a permanent visa	Need to physically present in Canada and have a permanent address when applying for the permit Cannot reside in Quebec Provide results of an approved English or French language test
Time toward permanent residency/citizenship	6 years (5 years for settled status and 1 extra year for citizenship)	PR: 3 years if the applicants live and work in designated regional areas, 4 years for the others	PR: 3 years
Potential beneficiary	5.4 million residents are eligible, and 300,000 residents expected to apply in the first 5 years	12,500 visa holders and 1250 applications on hand when the policy change was announced (under 9000 residing in Australia in November 2021)	All eligible applicants

BNO: British National Overseas.

In retaliation for the BNO visa program, the Hong Kong government declared the BNO passport an invalid travel document. This move, however, is more like a gesture than a deterrent for those who plan to take advantage of the special immigration program, because the issuing authority of the BNO passport is the British government, rather than the Hong Kong authorities. In addition, eligible Hong Kong citizens can still apply for the special immigration program without a physical copy of the BNO passport. Perhaps for this reason, the Hong Kong government took yet another step to bar Hong Kong citizens from moving to the United Kingdom. The Mandatory Provident Fund (MPF) Schemes Authority, a statutory body that oversees Hong Kong’s compulsory pension programs, announced on 10 March 2020 that BNO holders are not eligible for early withdrawal of their pensions. This decision likely impacts middle-aged workers more than the younger or older ones, because the former are old enough to accumulate substantial retirement savings in their MPF accounts and too young to withdraw the pensions for the retirement reason.

Despite the HKSAR government’s effort to deter emigration to the United Kingdom, many Hong Kong citizens managed to apply for the BNO visa. In the first 12 months after the launch of the immigration program, the British authorities received 103,900 BNO visa applications (Gilchrist, 2022), setting off a new wave of emigration. At the time of this writing, it is uncertain how many people will actually leave Hong Kong because this emigration wave is still ongoing. In all likelihood, however, its scale is no less significant than the emigration wave in the 1980s and the early 1990s. As shown in Figure 1, the net migration plunged to −1% of the Hong Kong population in 2020, markedly lower than any point in the city’s recent history.

Figure 1.

Net Movement as a Share of Hong Kong’s Population: 1980–2020.

Hypotheses

Post-2020 Hong Kong provides an excellent case to test our central argument, as the Hong Kong NSL has significantly elevated the cost of voicing out, while the special immigration programs that various countries offer to Hong Kong citizens has drastically reduced the exit cost. Loyalty here refers to the local identity. Based on the discussion in the previous section, we posit that

Hypothesis 1. Hong Kong citizens with a strong local identity are more interested in migrating to a foreign country.

We also argue that heterogeneous perceptions of performance decline play an important role in motivating loyal members to take the exit option. For this reason, we derive the following two hypotheses:

Hypothesis 2. Hong Kong citizens with a strong local identity tend to hold a more negative view of recent political changes.

Hypothesis 3. The negative view of recent political changes mediates the effect of local identity on migration intent.

The above three hypotheses denote, respectively, the direct and indirect effects of local identity on migration intent. Figure 2 presents a conceptual diagram of these effects.

Figure 2.

Direct and indirect effects of Hong Kong identity on migration intent.

Data and operationalization

Our data come from a pre-election survey of the 2021 Legislative Council election.⁵ We commissioned an international survey company to implement the Internet survey with its online panel. Using quota sampling to match the voting population based on gender, age, and districts, we obtained an effective sample of size N = 2024. The sample does not perfectly match the underling population. For this reason, we apply post-stratification weight to adjust for sampling variability in the following statistical analysis.

We operationalize the concept of loyalty using a survey question that asks how respondents would identify themselves. Five options are available: Chinese, Hong Kong Chinese, Chinese Hong Konger, Hong Konger, and others. We construct a variable of interest, Hong Konger, which takes a value of “1” if a respondent answers “Hong Konger” and “0” otherwise. For the original distribution of the five identity categories, see Figure 3. As may be seen from the figure, about half of the respondents identified themselves as a Hong Konger.

Figure 3.

Distribution of self-reported identity categories.

Three survey questions help us identify the outcome of interest. The first question asks respondents to report how frequently they discussed the idea of migration with friends and family in the past year. The second question is about the respondents’ willingness to migrate to a foreign country in the next 5 years, while the third asks respondents if they want their family members to move abroad.⁶ We also construct a new variable, Migration Intent, by extracting the latent factor of answers to the last two questions using principal component analysis. The first principal component accounts for 70% of the variance of the two variables.

Testing the causal mechanism involves identifying respondents’ perceived performance decline. The survey contains a number of questions asking respondents to evaluate the current situations of Hong Kong along several dimensions compared with 4 years ago. We focus on four dimensions that are subjected to the direct impacts of the Hong Kong NSL: civil liberties, judicial independence, electoral system, and educational system. We construct a single index, Perceived Political Changes, by extracting the latent factor of these four variables. The first component of our principal component analysis accounts for 81% of the variance, suggesting that these variables are highly correlated.

To reduce potential omitted variable bias, we adjust for a number of demographic and socioeconomic variables, including age, gender, income, education, and birthplace. As mentioned, a large number of Hong Kong citizens who hold the BNO passport can now easily migrate to the United Kingdom, thanks to the British government’s special immigration program. We also include a variable, BNO, that is assigned a value of “1” if the respondent reports holding a BNO passport or having a close relative who is a BNO holder and “0” otherwise. Wong and Wan (2018) point out that in Hong Kong wealth is a strong predictor of political attitudes in addition to income. For this reason, we also include a variable, Homeowner, which is assigned a value of “1” if the respondent owns either wholly or partially a property and “0” otherwise. We also control for respondents’ parental status, as L. Lui et al. (2021) show that Hong Kong parents tend to have stronger migration intent than non-parents. Finally, because political scientists have long identified economic performance is a key determinant of electoral outcomes (Lewis-Beck and Stegmaier, 2000), we constructed a composite index, Perceived Economic Changes, by extracting the latent factor of four survey questions related to the respondents’ evaluation of Hong Kong’s economic situation and their own economic status. Adjusting for this variable, we can ensure that the effect of Perceived Political Changes on Migration Intent is not confounded by perceived economic changes.

Using ordinary least squares (OLS), we regress Migration Intent on the variables of interest while adjusting for the demographic and socioeconomic variables.⁷ In some regression specifications, we interact Hong Kong Identity with BNO to examine whether being eligible for the BNO visa makes those who have a strong local identity even more prone to migration. In regressions where the outcome variable is not migration intent, our estimation strategy is ordered logistic regression (Table 2).

Table 2.

Local identity and migration intent.

Dependent variable	Migration intent		Frequency of family discussion of migration		Willingness to migrate: self		Willing to migrate: Family
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Hong Kong Identity	0.826**	0.736**	0.528**	0.538**	1.143**	1.024**	1.029**	0.904**
Hong Kong Identity	(0.070)	(0.103)	(0.099)	(0.145)	(0.100)	(0.148)	(0.102)	(0.150)
BNO	0.251**	0.178*	0.352**	0.360**	0.230*	0.135	0.324**	0.226
BNO	(0.063)	(0.083)	(0.096)	(0.132)	(0.093)	(0.126)	(0.093)	(0.122)
Hong Kong Identity × BNO		0.153		–0.017		0.201		0.212
Hong Kong Identity × BNO		(0.122)		(0.184)		(0.179)		(0.179)
Perceived political change	–0.165**	–0.165**	–0.139**	–0.139**	–0.226**	–0.226**	–0.252**	–0.252**
Perceived political change	(0.021)	(0.021)	(0.032)	(0.032)	(0.033)	(0.033)	(0.034)	(0.034)
Perceived economic change	0.040	0.039	0.049	0.049	0.038	0.038	0.032	0.032
Perceived economic change	(0.023)	(0.023)	(0.034)	(0.034)	(0.035)	(0.035)	(0.034)	(0.034)
Parent	0.294**	0.299**	0.444**	0.444**	0.336**	0.343**	0.373**	0.379**
Parent	(0.067)	(0.067)	(0.104)	(0.104)	(0.099)	(0.099)	(0.100)	(0.100)
Age	–0.022**	–0.022**	–0.016**	–0.016**	–0.036**	–0.036**	–0.026**	–0.026**
Age	(0.003)	(0.003)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)
Female	–0.053	–0.051	–0.033	–0.033	–0.180*	–0.177*	–0.027	–0.025
Female	(0.059)	(0.059)	(0.091)	(0.091)	(0.088)	(0.088)	(0.087)	(0.087)
Income	0.002	0.003	0.031*	0.031*	–0.007	–0.007	–0.010	–0.010
Income	(0.010)	(0.010)	(0.014)	(0.014)	(0.014)	(0.014)	(0.014)	(0.014)
Education	0.037	0.035	0.065*	0.065*	0.034	0.032	0.050	0.048
Education	(0.021)	(0.021)	(0.030)	(0.030)	(0.031)	(0.031)	(0.031)	(0.031)
Born in Hong Kong	–0.103	–0.093	–0.148	–0.149	0.003	0.016	–0.160	–0.146
Born in Hong Kong	(0.108)	(0.108)	(0.147)	(0.148)	(0.150)	(0.151)	(0.162)	(0.163)
Homeowner	0.026	0.022	0.044	0.044	–0.039	–0.044	–0.004	–0.010
Homeowner	(0.066)	(0.066)	(0.104)	(0.104)	(0.098)	(0.098)	(0.096)	(0.097)
Constant	0.301	0.338
Constant	(0.171)	(0.173)
Cutoff point 1			–0.762**	–0.758**	–2.773**	–2.820**	–2.263**	–2.313**
Cutoff point 1		(0.234)	(0.238)	(0.268)	(0.271)	(0.265)	(0.268)
Cutoff point 2			0.067	0.072	–0.842**	–0.891**	–0.395	–0.447
Cutoff point 2		(0.234)	(0.239)	(0.258)	(0.261)	(0.256)	(0.259)
Cutoff point 3			2.687**	2.691**	1.062**	1.017**	1.483**	1.436**
Cutoff point 3		(0.243)	(0.248)	(0.256)	(0.258)	(0.256)	(0.258)
Number of observations	1891	1891	1891	1891	1891	1891	1891	1891
R ²	0.24755	0.24822

BNO: British National Overseas.

The estimation strategy of the dependent variable Migration Intent is ordinary least squares (OLS) regression, while that of other dependent variables is ordered logistic regression. Standard errors are in parentheses.

p < 0.05; **p < 0.01.

Results

Local identity and migration intent

We first examine the relationship between local identity and migration intent.⁸ Table 1 displays the regression results. As may be seen from the table, Hong Kong Identity is a significant predictor at 1% in all specifications. Self-identified Hong Kongers are more likely to discuss migration issues with family and friends. They are also more likely to report willingness to move themselves or family members abroad within the next 5 years. The effect of Hong Kong Identity is also substantively significant. Consider Specification (2), where the outcome variable is the latent factor, Migration Intent. The coefficient on the variable of interest is 0.736. The effect size is fairly large because the difference accounts for about half of a standard deviation of the outcome variable. The result supports Hypothesis 1.

As for other variables, BNO holders or their direct family members indeed have stronger migration intent. The coefficient on the variable, BNO, is significant in all but one specification. As for the interaction term, Hong Kong Identity X BNO, its effect, based on Specification (2), is negative although not significantly different from zero. To visualize the joint impacts of Hong Kong Identity and BNO, we create a marginal plot based on the estimates of Specification (2). As may be seen from Figure 4, conditional on Hong Kong Identity, the BNO status does make a significant and positive impact on one’s migration intent, although the effect size is somewhat modest.

Figure 4.

Marginal effects of local identity and BNO status.

Of the demographic variables, only parent, age, and education show consistently significant effects. In particular, parents and educated respondents have a greater migration tendency. Older respondents are less interested in moving abroad. Perhaps not surprisingly, the coefficient on perceived political changes is negative and statistically significant, suggesting that those who view recent political changes in a positive light are less interested in migration.

Local identity and perceived performance decline

Our argument hinges on the claim that loyal members tend to be more sensitive to the performance decline of their affiliated group, partly because their demand for group-related information is stronger than their less loyal counterparts. In the context of Hong Kong, we expect that those who have a strong local identity would consume more local news than those with a weak identity. While the survey has no question on local news consumption, it does contain some questions related to respondents’ news consumption patterns on social media, which provide an indirect test of our claim. In particular, the three questions ask the respondents to report their frequency of sharing and discussing news on social media and the frequency of producing their own news commentaries on social media. The reported behavioral differences between Hong Kongers and non-Hong Kongers are shown in Table 3.

Table 3.

Differences in news consumption between self-identified Hong Kongers and others.

	(A) Hong Kongers		(B) Others		B − A	t statistic
	Mean	Standard error	Mean	Standard error	B − A	t statistic
Sharing news on social media	1.13	0.04	0.78	0.04	–0.35	–6.05
Discussing news on social media	1.23	0.04	0.81	0.04	–0.42	–7.31
Producing news commentaries on social media	0.60	0.03	0.40	0.03	–0.20	–4.42

As may be seen from the table, the news consumption patterns of these two groups are markedly different. Those who self-identify as Hong Kongers are more likely to share and discuss news on social media. They are also more active in producing news commentaries. Assuming that Hong Kong people are generally exposed to local news more than international news, we find prima facie evidence that local identity is associated with a greater demand for local information.

Next, we examine the difference in perceived performance decline between self-identified Hong Kongers and others. Figure 5 displays the distribution of the variable Perceived Political Changes by local identity. Consistent with our prediction, respondents with a strong local identity (denoted by the green area) hold decidedly negative views of recent political changes. Indeed, their distribution peaks at −2, which is almost the lowest value on the range of the variable. By contrast, for those who do not identify themselves as Hong Kongers (denoted by the red area),⁹ their distribution centers around 1, indicating that their views of recent political changes are much positive. Regressing the variable Perceived Political Changes on local identity and the same set of control variables as in Table 1, we find that the coefficient on the variable, Hong Kong Identity, is significantly negative with a t statistics at −18.45.¹⁰ The result supports Hypothesis 2: those who have a strong local identity perceive a dramatic performance decline compared with those who have a weak local identity.

Figure 5.

Perceived Political Changes by identity status.

Finally, we examine perceived political changes as a mediator between local identity and migration intent. We compute the average causal mediation effects (ACME) proposed by Imai et al. (2010), which refer to the indirect effect of the variable of interest (i.e. Hong Kong Identity) on the outcome (i.e. Migration Intent) via the potential mediator (i.e. Perceived Political Changes), holding all other factors constant.¹¹

Figure 6 displays the ACME estimate along with the estimates of the total and direct effects. As may be seen from the figure, all the effects are statistically distinguishable from zero. In particular, the ACME is about 0.22, suggesting that Perceived Political Changes is an important mediator. The result supports Hypothesis 3. It is also worth noting that local identity by itself also has a sizable direct effect on migration intent. In fact, the effect size of its direct effect is about 3.7 times of its indirect effect. An important implication is that there is something peculiar to the Hong Kong identity that drives people’s migration intent not through the channel of perceived performance decline.

Figure 6.

Mediating effect of perceived political changes.

Discussion

In this article, we use post-2020 Hong Kong as a case to show that loyalty to a group or a community actually encourages exit, which is at odds with Hirschman’s observation presented in his seminal book Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States. Our findings do not imply that Hirschman’s conception of human responses to decline is wrong. Rather, they highlight the danger of applying the EVL dynamics to a political context, for there is an important difference between firms and governments: the evaluation of the performance of latter is much more complicated than that of the former. The performance decline of an ice cream company manifests itself in the quality of the ice cream it produces. For governments, as Dowding et al. (2000) rightly point out, it is more difficult to detect performance decline because the services a government provides are multifaceted. This ambiguity, therefore, gives rise to heterogeneous perceptions of performance decline. In an authoritarian context, as we point out, “loyal” citizens—namely, those who strongly identify with the community—tend to be more sensitive to the performance change compared with the less “loyal.”

The findings of this article also provide implications for understanding Hong Kong politics. If migration intent is a significant predictor of migration decision, our results indicate that many who have left the city in the recent emigration wave, together with many more who will, are predominantly citizens with a strong local identity. These individuals are also likely dissatisfied with the political changes since 2019. Their departure will reduce and weaken the voices of dissent in Hong Kong, arguably a desirable scenario from Beijing’s point of view. Meanwhile, we expect to see Hong Kong diasporas to play an increasingly active role in resisting China’s overseas influences in many years to come, although the magnitude of their impacts remains unclear.

Last but not least, one thing that is missing in Hirschman’s theory is the decision to return (Hoffmann, 2010). In the politics of migration, it is entirely possible that outgoing migrants may one day go back to their origin. Indeed, Hong Kong in the late 1990s experienced a wave of returning migrants, who had moved abroad in the 1980s and the early 1990s and had chosen to return Hong Kong after acquiring a foreign passport and witnessing a relatively smooth sovereignty transfer. It is unclear if the Hong Kong emigrants of the current wave will at some point in the future relocate back to their hometown as British citizens. If they do, they are likely more politically vocal than the local, not only because they have a strong local identity, but also because their British citizenship allows them to exit from the city almost anytime they wish. Of course, equally likely is that these outgoing Hong Kong citizens may begin to identify with their new home country and detach themselves from Hong Kong affairs.

Footnotes

Appendix 1

Table 5.

Checking multicollinearity.

Variable	VIF	1/VIF
Age	1.36	0.73434
Hong Kong Identity	1.34	0.749019
Perceived Political Changes	1.33	0.750655
Parent	1.28	0.779516
Homeowner	1.27	0.789139
Income	1.19	0.8376
Education	1.14	0.877256
BNO	1.1	0.905873
Born in Hong Kong	1.08	0.925122
Female	1.02	0.977445

BNO: British National Overseas; VIF: variance inflation factor.

The VIFs are calculated based on Specification (1) of Table 1.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: In this study we used the public opinion data of the Hong Kong Election Study, which received funding support from the University Grants Committee of Hong Kong (GRF No.: 15605519)

ORCID iD

Stan Hok-Wui Wong

Notes

Author biographies

Stan Hok-Wui Wong is a Visiting Associate Professor at the University of St. Thomas. He is the principal investigator of the Hong Kong Election Study. His works have been published in journals such as the British Journal of Political Science, China Quarterly, Electoral Studies, Journal of Peace Research, and Political Behavior.

Kelvin Chun-Man Chan is a PhD student at the School of Economic, Political, and Policy Sciences at the University of Texas at Dallas. His research interests include electoral politics, regional cooperation, and foreign policies of China and Japan.

Karl Ho is Associate Professor of Instruction in the School of Economic, Political and Policy Sciences at the University of Texas at Dallas. His research and teaching areas cover elections, public policy and political economy with a regional focus on East Asia. He is the co-Principal Investigator of the Hong Kong Election Study project, collecting election survey data since 2015.

The late Harold D. Clarke was the Ashbel Smith Professor in the University of Texas at Dallas. He has served as editor of Electoral Studies and the Political Research Quarterly and as Director of Social and Economic Sciences at the National Science Foundation. His research has been supported by the National Science Foundation (US), the Economics and Social Research Council (UK), the Social Sciences and Humanities Research Council of Canada, the Canada Council, the Bill and Melinda Gates Foundation and the Hong Kong Science Foundation.

References

Amira

Wright

Goya-Tocchetto

(2021) In-group love versus out-group hate: which is more important to partisans and when? Political Behavior 43(2): 473–494.

Appiah

Knobloch-Westerwick

Alter

(2013) Ingroup favoritism and outgroup derogation: effects of news valence, character race, and recipient race on selective news reading. Journal of Communication 63(3): 517–534.

Barry

(1974) Review article: “exit, voice, and loyalty.” British Journal of Political Science 4(1): 79–107.

Bennett

Durkin

(2000) The effects of organisational change on employee psychological attachment: an exploratory study. Journal of Managerial Psychology 15(2): 126–147.

Berry

(1997) Immigration, acculturation, and adaptation. Applied Psychology 46(1): 5–34.

Birch

(1975) Economic models in political science: the case of “exit voice, and loyalty.” British Journal of Political Science 5(1): 69–82.

Bowlby

(1973) Attachment and Loss: Separation, Anxiety and Anger, vol. 2. London: The Hogarth Press and The Institute of Psycho-Analysis.

Bowlby

(1980) Attachment and Loss: Loss, Sadness and Depression, vol. 3. London: The Hogarth Press and The Institute of Psycho-Analysis.

Colic Peisker

Walker

(2003) Human capital, acculturation and social identity: Bosnian refugees in Australia. Journal of Community & Applied Social Psychology 13(5): 337–360.

10.

Davis-Blake

Broschak

George

(2003) Happy together? How using nonstandard workers affects exit, voice, and loyalty among standard employees. Academy of Management Journal 46(4): 475–485.

11.

Dickson

(2016) The Dictator’s Dilemma: The Chinese Communist Party’s Strategy for Survival. Oxford: Oxford University Press.

12.

Dowding

John

Mergoupis

, et al. (2000) Exit, voice and loyalty: analytic and empirical developments. European Journal of Political Research 37(4): 469–495.

13.

Duque

Lado

(2010) Cross-cultural comparisons of consumer satisfaction ratings: a perspective from Albert Hirschman’s theory. International Marketing Review 27: 676–693.

14.

Ein-Dor

Hirschberger

(2016) Rethinking attachment theory: from a theory of relationships to a theory of individual and group survival. Current Directions in Psychological Science 25(4): 223–227.

15.

Gehlbach

(2006) A formal model of exit and voice. Rationality and Society 18(4): 395–418.

16.

Gilchrist

(2022) ‘Hong Kong is not going to be under the rule of law’: more than 100,000 apply for new visa to Britain. CNBC, 3 March. Available at: https://www.cnbc.com/2022/03/02/hong-kong-bno-visa-100000-apply-to-live-in-united-kingdom.html (accessed 3 May 2022).

17.

Hirschman

(1970) Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States. Cambridge, MA: Harvard University Press.

18.

HKSAR Census and Statistics Department (2022) Table 2: population growth by component. Available at: https://www.censtatd.gov.hk/en/web_table.html?id=2 (accessed 6 May 2022).

19.

HKUPOP (2019) People’s ethnic identity. Available at: https://www.hkupop.hku.hk/english/popexpress/ethnic/eidentity/poll/datatables.html (accessed 3 May 2022).

20.

Hoffmann

(2010) Bringing Hirschman back in: “exit,” “voice,” and “loyalty” in the politics of transnational migration. The Latin Americanist 54(2): 57–73.

21.

Hoffner

Rehkoff

(2011) Young voters’ responses to the 2004 US presidential election: social identity, perceived media influence, and behavioral outcomes. Journal of Communication 61(4): 732–757.

22.

Imai

Keele

Yamamoto

(2010) Identification, inference and sensitivity analysis for causal mediation effects. Statistical Science 25(1): 51–71.

23.

Kaeding

(2017) The rise of “localism” in Hong Kong. Journal of Democracy 28(1): 157–171.

24.

Knobloch-Westerwick

Hastall

(2010) Please your self: social identity effects on selective exposure to news about in-and out-groups. Journal of Communication 60(3): 515–535.

25.

(2004) Immigration policies, discourses, and the politics of local belonging in Hong Kong (1950–1980). Modern China 30(3): 326–360.

26.

Kucuk

(2008) Consumer exit, voice, and “power” on the Internet. Journal of Research for Consumers. Available at: http://jrconsumers.com/academic_articles/issue_15/exit-voice-power-academic2.pdf

27.

Lau

Kuan

(1988) The Ethos of the Hong Kong Chinese. Hong Kong, Hong Kong: Chinese University Press.

28.

Lee

Yuen

Tang

, et al. (2019) Hong Kong’s summer of uprising. China Review 19(4): 1–32.

29.

Lee

Whitford

(2008) Exit, voice, loyalty, and pay: evidence from the public workforce. Journal of Public Administration Research and Theory 18(4): 647–671.

30.

Leung

(2019) 16AUG2019 yingmeigangmeng zhuquan zaimin jihui (Liang jiping yueyang fayan) [16AUG2019 Stand With Hong Kong, Power to the People Rally (Brian Leung Kai-ping’s overseas speech)] [Video]. SocREC, 17 August. Available at: https://www.youtube.com/watch?v=pC57wbg3z8Y

31.

Lewis-Beck

Stegmaier

(2000) Economic determinants of electoral outcomes. Annual Review of Political Science 3(1): 183–219.

32.

Lui

Sun

KCY

Hsiao

(2021) How families affect aspirational migration amidst political insecurity: the case of Hong Kong. Population, Space and Place, 28: e2528.

33.

Lui

Chiu

(1999) Social movements and public discourse on politics. In Ngo

(Ed.), Hong Kong’s history : state and society under colonial rule (pp. 101–118). London; New York : Routledge.

34.

(2015) The rise of “anti-China” sentiments in Hong Kong and the 2012 Legislative Council elections. The China Review 15: 39–66.

35.

Meyer

Herscovitch

(2001) Commitment in the workplace: toward a general model. Human Resource Management Review 11(3): 299–326.

36.

Pfaff

Kim

(2003) Exit-voice dynamics in collective action: an analysis of emigration and protest in the East German revolution. American Journal of Sociology 109(2): 401–444.

37.

Power

(2017) Discerning devotion: testing the signaling theory of religion. Evolution and Human Behavior 38(1): 82–91.

38.

Rusbult

Zembrodt

Gunn

(1982) Exit, voice, loyalty, and neglect: responses to dissatisfaction in romantic involvements. Journal of Personality and Social Psychology 43(6): 1230–1242.

39.

Sosis

Ruffle

(2001) Religious Ritual and Cooperation: Testing for a Relationship. Tel-Aviv, Israel: Pinhas Sapir Center for Development, Tel-Aviv University.

40.

Stein

Hobson

Schroeder

(2021) A sacred commitment: how rituals promote group survival. Current Opinion in Psychology 40: 114–120.

41.

Veg

(2017) The rise of “localism” and civic identity in post-handover Hong Kong: questioning the Chinese nation-state. The China Quarterly 230: 323–347.

42.

Wong

SHW

(2012) Authoritarian co-optation in the age of globalisation: evidence from Hong Kong. Journal of Contemporary Asia 42(2): 182–209.

43.

Wong

SHW

Chan

KCM

(2021) The impact of pre-election protests in electoral autocracies: the case of Hong Kong’s Anti-ELAB Movement. Issues & Studies 57(3): 2150010.

44.

Wong

SHW

Wan

(2018) The housing boom and the rise of localism in Hong Kong: evidence from the legislative council election in 2016. China Perspectives 2018(2018/3): 31–40.

45.

Wong

(1988) Emigrant Entrepreneurs: Shanghai Industrialists in Hong Kong. Hong Kong, Hong Kong: Oxford University Press.

46.

Yuen

Cheng

(2017) Neither repression nor concession? A regime’s attrition against mass protests. Political Studies 65(3): 611–630.