Regional informal institutions, local governance and internal migration in Vietnam

ABSTRACT Internal migration is an institutionalized process; it is affected not only by economic factors but also by institutional forces. This study thus employs institutional theory to investigate the importance of regional informal institutions and local governance quality on types of migration, as well as their impacts on migrants’ earnings. Examining a set of national internal migration data in Vietnam, we find that pure socialist regions (the north) attract in-migration for job-seeking purposes, while pre-capitalist regions (the south) attract in-migration for business opportunity-seeking purposes. More importantly, migration across regions with dissimilar informal institutions yields adverse effects on migrants’ earnings. Finally, migration from regions with a lower governance quality to regions with a higher governance quality improves migrants’ earnings.


INTRODUCTION
Conventional models have investigated migration as an economic decision (Ravenstein, 1885(Ravenstein, , 1889Harris & Todaro, 1970;Sjaastad, 1962). However, recent empirical literature shows that migration cannot be fully explained by economic factors alone. For example, Carlsen and Leknes (2021) use a Rosen-Roback model and Bernini and Tampieri (2019) use a multilevel model to analyse more than 2.5 million Norwegians and 31,680 Italians, respectively. They show that migrants in these countries are more likely to be attracted to regions that have a higher level of subjective well-being conditions than regions that simply have better economic conditions. Also, Luong (2018), in the context of Vietnam, finds that many rural-urban migrants have returned home in the last few years, despite the higher earnings in big cities. These findings suggest that non-economic factors may play a role in shaping the flows of internal migration.
Built on these premises, we propose that institutions are an important determinant of migration. Following Williamson (2000), we adopt a framework of three-level hierarchical institutions, in which informal institutions (norms, values and beliefs) are positioned at the highest level, then formal institutions (laws, regulations and constitutional configurations) come at the second, and local institutions of governance (i.e., the execution of formal institutions) are at the third level. Given that this is a within-country study (Vietnam being the context of analysis), national formal institutions are identically applied to all regions; however, there are regional differences in the informal institutions and institutions of governance.
First, in terms of informal institutions, during the Second World War, Vietnam was divided into two states that have distinct institutional settings. While the north of Vietnam followed the socialist blueprint from the outset, the south of Vietnam was exposed to capitalism during the period 1954-75. As such, individualism and pro-entrepreneurship are the main values of people in the south, whereas collectivism and socialism are valued by people in the north . This difference in regional informal institutions is sticky, holding true even after 45 years of reunion (Dell et al., 2018;Nguyen, 2021). As such, it is important to understand how the differences in informal institutions between the north and south of Vietnam influence migration and, more importantly, whether crossregion migration improves or impair migrants' earnings.
Second, in terms of institutions of governance, the quality of local government across Vietnam's regions is also inconsistent. This is because the formal institutional frameworks in developing countries are incomplete and underdeveloped (Du & Mickiewicz, 2016), leading to a situation in which local authorities are required to implement their own interpretations of the rules (Malesky et al., 2015). This results in significant regional variations in how national policies and regulations are executed (Nguyen, 2020). Hence, it is likely that the differences in local governance quality across Vietnam's regions will influence migration. In this study, we therefore explore whether it is beneficial for migrants' earnings to move from regions with a lower quality of governance to those with a higher quality of governance.
To answer these questions, we examine the Vietnam National Internal Migration (VNIM) dataset, which comprises 8009 non-migrants and 8961 migrants in Vietnam in 2004 and 2015. To the best of our knowledge, this is the most comprehensive and representative data on internal migration in Vietnam. Unlike previous studies that focus solely on job-seeking migrants, we aim to distinguish the effects of institutions on two types of migration, that is, migration for job-seeking purposes versus migration for business opportunity-seeking purposes. This framework thus extends previous research out of the conventional model examining rural-urban migrants looking for low-skilled employment opportunities (Haughton et al., 2018;Nguyen et al., 2017b).
Our findings suggest that migrants in the north of Vietnam (where collectivism and socialism are the dominant values) are more likely to move in the region for job-seeking purposes; however, in the south of Vietnam (where individualism and pro-entrepreneurship are the primary values), migrants are more likely to move in for business opportunity-seeking purposes. Also, it is found that migration across the north-south regions is associated with reduced earnings. In terms of local governance, regions with higher quality of governance attract migrants coming for both job-seeking and business opportunityseeking purposes. Finally, moving to a region that has a higher governance quality improves migrants' earnings.
This study contributes to the regional study literature in several ways. First, it explores the differences in types of migration across regions that have distinct informal institutions and shows that migration across these regions is not financially beneficial to migrants. Second, this study highlights the importance of local governance to migration. Specifically, it shows that in a country that has a single formal institutional framework, migrants are sensitive to the surrounding governance arrangements when they are making their migration decisions. In general, this study proposes that within a country, non-formal institutional forces, such as informal institutions and local governance, should be factored into the policymaking process to enhance the efficiency of labour force allocation.

LITERATURE
2.1. Institutional theory Institutions shape the context in which individuals behave and operate, and therefore affect both the opportunities and risks they face (Tolbert & Zucker, 1996;Carpenter et al., 2006). Specifically, North (1990) argues that many of the incentives underlying social and economic decisions are a function of institutional quality. He proposes a twopillar framework of formal and informal institutions, in which formal institutions are the explicit rules (e.g., constitutional laws and national legal systems) while informal institutions are the implicit rules (e.g., customs, traditions and social norms) that a society accepts as its standards of behaviours and which drive individuals' decision-making.
Delving more deeply into the structure of institutions, Williamson (2000) proposes a three-level hierarchical framework, placing informal institutionssocial embeddednessat the most profound position in the institutional structure. The reason for this is that these implicit institutional forces are the deepest rooted and slowest to change (Fritsch & Mueller, 2007;Fritsch & Mueller, 2008). In the context of Vietnam, following Dell et al. (2018) and Nguyen (2021), we propose that the differences in values, norms and beliefs between the north and south of Vietnam are deeply societal embedded and remain unchanged even after the two states unified.
Formal institutions are located at the second level. For example, in the context of Vietnam and China, the Hokhau (Hukou) system of resident registration is the institutional force that has been most commonly investigated in relation to migration decisions Zhai & Wang, 2002). Under this system, each household is given a registration booklet that records the name, sex, date of birth, marital status and occupation of each household member and the relationship to the household head . It is used for the purposes of identification, access to subsidized healthcare, education and some social benefits, as well as to restrict migration into the cities (Andrew, 2001). Williamson's (2000) third level of institutions is governance. This level is concerned with how the formal institutions are executed and implemented at the local level. This institutional dimension is particularly important to migration decisions, especially in developing countries, for two reasons. First, the formal institutional frameworks in developing countries are incomplete and underdeveloped; as such, they may not be fully, consistently, and efficiently executed across a country's regions, but are likely to vary according to how they are interpreted and enforced by local governments (Nguyen, 2020;Malesky et al., 2015). This creates a situation in which there is substantial heterogeneity in local governance quality across regions that apparently have an identical set of national formal institutions. Second, people who are working and living in a particular region are more sensitive to their surrounding institutional environments (e.g., local governance arrangements) than they are to the very broad national general constitutional configurations (Vu & Jytte, 2012).
Unfortunately, while formal institutions have been shown to influence migration decisions and earnings (Stark & Lucas, 1988;Li & Zahniser, 2002), the role of informal institutions and institutions of governance remain largely unknown. In this study, which holds national formal institutions constant (i.e., it is a withincountry study), we examine the influence of regional informal institutions and local governance quality on migration decisions, and the associated effects of these on migrants' earnings. Figure 1 illustrates the analytical framework, which is adapted from the model of Williamson (2000) to our research topic.

Regional informal institutions and institutions of governance in Vietnam
The differences in informal institutions between the north and south could be traced back to the development of Vietnam. Specifically, the root of the modern Vietnam is Daivieta small nation located in the north-west area, which was governed by China in most of its 2000-year history (Taylor, 1998). In 938, Daiviet gained independence from China and expanded its territories to the south. Despite being independent, its political organization was heavily influenced by the Chinese statecraft and the associated informal institutions, including collective action norms. Dell et al. (2018) explain that since the village was the foundational administrative unit in the Chinese statecraft for collecting taxes, villagers needed to work collectively and help each other to ensure that their villages, as a whole, met the tax requirements of the state. This format of social interactions gradually nurtured collective action norms and communal values (Nguyen, 2021).
Meanwhile, the south of Vietnam was following the Hindu-Buddhist statecraft imported from India and Cambodia (Banerjee & Iyer, 2002). In this system, power was organized by levels, peasants paid tribute to their landowning patrons; these patrons, in turn, had their own relations with higher level patrons (Acharya, 2013). This decentralized and personalized patron-client model thus facilitated individualism-oriented values in the south. It is noteworthy that the southernmost section of Vietnam was merged into Daiviet in 1698, a mere 150 years before the French colonization (Dell et al., 2018), allowing the two regions to sustain their distinct informal values and norms.
Interestingly, the Second World War further expanded the differences associated with the two regions.
Specifically, Vietnam was divided into two states that followed distinct institutional settings: North Vietnam and South Vietnam. The former followed the socialist blueprint, whereas South Vietnam was exposed to capitalism over the period 1954-75. During partition, the authorities in North Vietnam spread socialist values and planned the economy by, inter alia, nationalizing enterprises, illegalizing private properties/private businesses, and discriminating against entrepreneurship. Meanwhile, they boosted values such as collectivism by legitimizing and promoting collective action norms, such as Hoptacxa, which is a form of cooperative business created by groups of farmers that has legal standing (Lei & Chen, 2011;Dell et al., 2018).
During the same period, pro-Western authorities in South Vietnam were boosting capitalist values and governance systems, encouraging international trades, private businesses and entrepreneurship . They also promoted their own set of values, such as individualism, independence and self-reliance by encouraging the citizenry to act in pursuance of their goals and desires (Dell et al., 2018). The individualistic norms that originated from South Vietnam's exposure to capitalism, and which contrast with the collectivist norms of North Vietnam, were retained after the two states reunified (Nguyen, 2021;, in which respect Vietnam is similar to East and West Germany (Fritsch & Mueller, 2008;Fritsch & Wyrwich, 2014).
Appendix A in the supplemental data online provides some economic indicators showing that the differences between the north and south of Vietnam remain until today. The statistics are calculated using the data published on the Vietnam General Statistics Office website. 1 It can be seen that the south has more private businesses per 1000 citizens than the north. In terms of attracting foreign direct investment (FDI), the south shows a clear advantage, in which it has 4746 projects and almost US $8 billion more than the north by the end of 2020. In terms of average incomes, the south consistently has a higher level than the north. The better economic performance of the south is attributed to the pro-entrepreneurial values and norms embedded in the local community and have been consistently found in different periods of time, Figure 1. Analytical framework.
Regional informal institutions, local governance and internal migration in Vietnam such as three decades ago (Freeman, 1996), two decades ago (Kim, 2007) and recently .
Not only do the informal institutions (norms, values and beliefs) vary across Vietnam's regions, local governance quality also significantly differs. These local governance variations result from the extensive decentralization during the Doimoi (economic reform) (Lan Phi & Anwar, 2011). This programme was founded on the promulgation of the 1996 (revised in 1998) State Budget Law, which granted local governments considerable autonomy in their fiscal strategies. Subsequently, local authorities were given substantial independence in determining their local governance structures and regulatory arrangements (Nguyen, 2020). These variations in local governance arrangements, which lead to substantial differences in economic growth among regions, gave rise to an increasing wave of internal migration across the nation, despite the restrictions imposed by Hokhau . For example, Vu and Jytte (2012) find that Vietnamese wishing to move out of their hometowns actively develop their social migration networks over time; the built-up social capital helps them circumvent the state control over population mobility and thus allow them to relocate.

Migration theory
Migration as an economic activity was first theorized by Ravenstein (1885Ravenstein ( , 1889. His 'laws of migration' assert that both pull and push factors are essential to the decision to migrate. The pull factors are the (usually good) socioeconomic and political environments of the destinations that attract people from other regions to in-migrate. Meanwhile, the push factors are the (usually bad) living and working conditions at the places of origin that force people to out-migrate. Ravenstein positions migration as a purely economic decision and he focuses on economically determining factors such as the development of industry and commerce, transportation infrastructure, and labour force surpluses and deficits. Meanwhile, institutional determinants, including local governance quality and informal institutions, are largely ignored. As such, even though his works have laid the foundation for the classical migration models (Sjaastad, 1962;Harris & Todaro, 1970), empirical evidence shows that this strand of theorizing on migration is not always valid.
For example, Chen and Coulson (2002) examine a sample of 223 Chinese cities and find that the attributes of the quality of urban life, such as housing-market conditions and transport, have little explanatory power on rural-urban migration decisions. Also, Luong (2018) find that numerous rural-urban migrants in Vietnam have returned home in the last few years, despite the higher incomes they earned in big cities. He argues that this changing configuration has to do with a socio-culturally constructed web of moral obligations to family members in some regions. These findings suggest that migration is not a purely economic decision but is also a function of institutional forces.

Migration in Vietnam
In Vietnam, migration has been examined in association with urbanization and industrialization. According to the Internal Migration Booklet, in 2015, 19.7% people living in urban areas were migrants, while the corresponding number in rural areas was 13.4%. The south-east had the highest level of migration, 29.3%. Moreover, 79.1% migrants originally came from the rural while 20.9% were from the urban in 2015. Interestingly, 52.4% migrants were females, confirming the feminizing trend of migrants in developing countries (Tacoli & Mabala, 2010). Migrants' incomes were (5 million VND), on average, lower than those of non-migrants (5.4 million VND). In a comparison between the north and south. Summary statistics in Table 1 (column 9) show that the north has more migrants coming in for job opportunities; meanwhile, the south has more migrant entrepreneurs. It could also be seen that, in terms of cross-region migration, the main direction flow is north to south. This could be explained by the fact that the south hosts the largest economic hub: Hochiminh city and its surrounding industrial zones (Liu, 2019). In terms of education, the average level in the north is higher than it is in the south. However, when it comes to aspirations to increase incomes, people in the south show a stronger incentive to do so. Finally, it is also observed that migrants in the north are less likely to move out of their home provinces compared with migrants in the south. This might signal that people in the south are keener on taking risks in search for better opportunities (Truong & Schuler, 2021).
It is noteworthy that, in the context of Vietnam, most research has been done for the group of migrants moving for employment, setting aside the other group of migrants moving for business opportunities. For example, a large body of the literature has focused on investigating rural-urban migration decisions of employees and the subsequent impacts on their earnings and living standards (Liu, 2019;Haughton et al., 2018;Nguyen et al., 2017a). Prior research shows that rural-urban migration for employment helps reduce poverty and inequality, and improves migrant households' consumption of education and healthcare services (Nguyen et al., 2015;Amare & Hohfeld, 2016). An implicit assumption in these studies is that the positive effects of internal migration are homogenous across regions within Vietnam. This may not be valid, unfortunately, when we consider the heterogeneity of the subnational institutional forces, including regional informal institutions and local governance quality.
For example, Lanjouw and Marra (2018) use data from the Vietnam Population and Household Census to show that migrants to small towns typically encounter public services that are of much lower quality than those enjoyed by migrants to large cities. The authors suggest that policymakers concerned with rural-urban migration should not neglect the institutional quality of smaller towns. In another study, Fesselmeyer and Le (2010) argue that in Vietnam government policies have created some benefit to urban dwellers at the expense of rural areas, lending  (Lipton, 1989), which states that government, under strong political pressure from the urban population, directs resources from rural to urban areas without consideration of efficiency or equity. These studies thus highlight the heterogeneity in the development of institutional environments in the context of Vietnam.
In terms of informal institutions, Vu and Jytte (2012) adopt a historical perspective to explain how migrants resist and undermine state control over population mobility in order to affect their movement and migration. The authors' findings imply that informal institutions (such as informal networks of migrants) can help migrants overcome the constraints of formal institutions like Hokhau system. Unfortunately, the study does not show the differences in these informal institutions across regions, which are likely to play a determining role.
In short, the extant empirical literature in Vietnam lacks a systematic analysis of the importance of institutional forces in explaining types of migration and their subsequent impacts on migrants' earnings. We will address these gaps in this study.

HYPOTHESES
Built on the above literature, in this section, we propose a set of hypotheses that link (1) regional informal institutions and (2) local governance quality to types of migration and their effects on migrants' earnings.
3.1. Institutions and types of migration 3.1.1. Informal institutions and types of migration First, we propose that migrants moving for job-seeking are more likely to come to the north of Vietnam. Meanwhile, migrants moving with an aim of searching for business opportunities are more likely to come to the south of Vietnam. This division of migration flows based on geographical locations is built on the insight of the institutional theory. Specifically, since the north of Vietnam follows socialist ideologies, firms located in this region are more likely to employ the collective and common distribution of rewards to all employees, which may not strongly correlate with contributions (Truong & Schuler, 2021;O'Toole et al., 2016). Also, they employ tenure employment schemes in which employees, when being hired, cannot be fired easily regardless of their performance (Nguyen & van Dijk, 2012). Moreover, the north of Vietnam locates most of the nation's state-owned enterprises (SOEs) (Nguyen & van Dijk, 2012), whose offered jobs are relatively less demanding while the opportunities for employees to obtain informal/extra rewards are high, due to the privileges and special treatments these firms obtain from the governments (Minor et al., 2018).
The attractiveness of the north to migrant employees could also be explained through the lens of social capital theory. Social capital is defined as a person's social characteristics that enable him or her to reap returns from interactions with members in his or her networks (Glaeser et al., 2002). Using this concept, Boenisch and Schneider (2013) build on the seminal works of Granovetter (1973) and Putnam (1993) to propose a framework of open versus closed types of social capital, in which the open type represents social networks that are built on formal and weak ties such as professional associations; meanwhile the closed types indicate social networks that are built on informal and strong ties such as cliques and friends. Using the case of East and West Germany, they demonstrate that people in the socialism region (the East) are stocked Borrowing this framework, we suggest that people in the north of Vietnam are more likely to build quasi-kinship relations with members in their networks (Dell et al., 2018), including migrants. This allows migrants to obtain personal help and support, which lower their relocation costs (time and effort). As such, migrant employees in the north may find it easier to set up strong and informal networks with their colleagues and managers, potentially leading to higher earnings. Evidently, Truong and Schuler (2021) find that north Vietnamese are more open to migrants than south Vietnamese. This type of social capital may be helpful to migrant employees but may not be conducive to migrant entrepreneurs, whose activities are transaction-based, which require formal, arm's-length relations to reduce transaction costs. As such, the type of social capital in the north may attract migrant employees more than migrant entrepreneurs.
Meanwhile, in the south of Vietnam, entrepreneurial values stemming from capitalism remain essential in the business community . The society highlights innovation, productivity and performance-driven behaviours (Truong & Schuler, 2021). Entrepreneurs are thus more likely to find social support for their business ideas in the south of Vietnam. For example, Nguyen (2021) shows that the capitalist norms and values in the south of Vietnam help boost the number of newly created businesses significantly higher than it is in the north. As such, it could be expected that the region attracts more migrants moving in for business opportunity-seeking purposes.
Moreover, from the lens of social capital theory, the open (i.e., formal and weak ties) social capital in the south provide entrepreneurs with more information required to run business. Granovetter (1973) and Burt and Burzynska (2017) emphasize the strength of weak ties in accessing new information and resources. Strong and informal ties may be helpful to migrants (including entrepreneurs and employees) in settling but this type of social networks does not facilitate entrepreneurship since it is typically associated with a structure of many small and isolated groups, which reduce information flows and increase transaction costs (Boenisch & Schneider, 2013). As such, we propose the following: Hypothesis 1: Migrants moving for job-seeking purposes are more likely to come to the north of Vietnam (pure socialism). Migrants moving for business opportunity-seeking purposes are more likely to come to the south of Vietnam (formerly capitalism).
3.1.2. Institutions of governance and types of migration Besides regional informal values, local governance quality also influences migration. Governance quality indicates the effectiveness of local authorities in executing national laws and/or initiating their own regulations and policies in improving their citizens' welfare and living standards (Malesky et al., 2015). Local governance quality is a multidimensional construct . In this study, we propose that the following six dimensions of governance matter to migrants' moving decisions. First, participation at local levels, which indicates the opportunities for citizens to engage in the public decision-making process as well as the quality of local elections, will influence the perception of migrants about the level of democracy in the regions. Barsbai et al. (2017) show that migrants prefer to move to regions/countries which are more democratic. The second governance factor is accountability, which indicates the level of interactions with local authorities and governments' response to citizens' appeals. Migrants prefer moving to regions/countries with a higher level of accountability since their voices are more likely to be heard (Cebula, 1974).
The third and fourth dimensions are transparency and control of corruption. These two factors seem similar but are distinct by nature. Transparency denotes access to information and communal budget and expenditure while control of corruption indicates the willingness of local authorities to fight against corruption and corresponding measures employed to limit bribery in public service delivery (Malesky et al., 2015). These two governance dimensions affect the costs of settling as such influence migrants' moving decisions (Poprawe, 2015). The fifth and sixth factors are administration procedures and public services. The former indicates the level of bureaucracy associated with certification procedures, including the resident registration Hokhau; meanwhile, the latter denotes the quality of public services such as infrastructure, law and order, health and education. Unsurprisingly, migrants want to move to regions/countries with more efficient administration procedures and better public services to improve their quality of life (Ogunniyi et al., 2020).
In short, we argue that a region with a strong (more effective) governance system, as a combination of the six dimensions presented above, is more likely to attract migrants since it provides them with better living standards, amenities, and quality of life. As such, we have: Hypothesis 2: Migrants moving for job-seeking and business opportunity-seeking purposes are more likely to come to regions with a higher (more effective) local governance quality.

Institutions, types of migration and earnings 3.2.1. Informal institutions and migrants' earnings
Migration decisions inevitably affect migrants' earnings in their new home locations. This is particularly the case when migrants move across regions sharing dissimilar informal institutions. People moving from the north to the south of Vietnam may find socialist ideologies are less popular in local business norms (Dell et al., 2018). Meanwhile, people moving from the south to the north of Vietnam may feel the working/business environment less active and more tokenist (Truong & Schuler, 2021). These may create substantial difficulties for them to learn, accept and 'play' the new 'rules of the game', leading to reduced productivity and efficiency. Unfortunately, these costs of adaption may not be the sole reason that cut their earnings.
The mismatch in social capital between the two regions could be an additional explanation. Boenisch and Schneider (2013), in the context of Germany, argue that informal relationships created under the socialist regime are single-purpose provision networks intended to solve specific problems which are designed to adapt to an economy of shortage. Those kinds of networks will become useless in a market economy. On the other side of the coin, formal and strong social networks created in a proentrepreneurship regime will also reduce their efficiency in a socialist economy. This argument may apply to the north and south of Vietnam. People who are familiar with extracting capital (resources) from strong-tie, informal networks in the north may find they are resource-constrained when moving to the south; meanwhile, people who used to 'play' the arm's-length, formal 'game' of social networks may find they need to spend more time to invest in building family-alike relationships in the north. The process of adapting and transforming social capital takes time and effort (Putnam, 1993). With a lower level of social capital, migrants moving across the north-south regions may find their activities more constrained, leading to reduced earnings as a consequence.
On top of this, discrimination may also play a role. For instance, the taste-based theory (Becker, 1971) suggests that economic agents are keen to seek transaction partners who are similar to them, with the purpose of reducing uncertainties and improving perceived trust. When the transaction partners (e.g., employees) are from other tribes (e.g., from regions with dissimilar social values and beliefs), employers may be exposed to uncertainties, leading to discrimination. Also, information-based theory (Edmund, 1972) argues that employers are inclined to believe that migrant employees coming from regions with unfamiliar local norms and informal institutions are less trustable because of informational asymmetries (Ibarra, 1992). Similar logics applied to entrepreneurs migrating across the two regions as well. Since they are from another tribe, it will be more difficult for them to be socially accepted, making it more challenging for them to build trust and to show their capacity to local business partners, customers, and suppliers (Nguyen, 2021). Therefore, we propose: Hypothesis 3: Migrants moving across the north-south regions of Vietnam earn less than migrants moving within the north region or within the south region.

Institutions of governance and types of migration
Finally, we also examine the link between migrants' earnings and local governance quality. Being consistent with the institutional theory, it is expected that migrants moving from regions that have a lower governance quality to regions that have a higher governance quality find their earnings higher than those (whose share the same characteristics) that do not do so. This is because of the well-structured and efficiently executed governance systems and policies that allow local markets to function properly (Fritsch & Mueller, 2008). Under such a conducive environment, both entrepreneurs and employees can keep focused on their jobs with least distractions from government interventions (Du & Mickiewicz, 2016). Also, they may gain access to subsidies and supports such as skill training and business matchmaking services, leading to improved productivity, economic performance (Nguyen, 2020), and earnings as a consequence.
It is noteworthy that, according to the Schumpeterian viewpoint, creative destructions and their associated business opportunities are more likely to occur in a chaotic environment (e.g., war, dereliction and economic crises) than it is in a well-ordered environment (e.g., efficient governance) (Schumpeter, 1942). Nevertheless, we believe that, in the context of our study, migrants moving from a region having weaker governance to a region having a stronger one can improve their earnings, for two reasons. First, an improvement in institutional quality helps set the 'rules of the game' and create a clear societal order and incentive structure that encourages entrepreneurial activities (Malesky et al., 2015). In a less stable environment, property rights are not assured , thereby entrepreneurs may show less incentive to move for business-seeking purposes. Second, transaction costs (i.e., the costs of finding, negotiating and monitoring business partners) are lower in regions that have a higher quality of governance. This helps facilitate entrepreneurial activities, save time and money (e.g., bribery) for entrepreneurs. As such, the number of transactions conducted may increase, leading to the establishment of a more active local business community. This ultimately boosts both entrepreneurs' and employees' incomes as a result. As such, we propose: Hypothesis 4: Migrants moving from regions with a lower local governance quality to regions with a higher local governance quality earn more than migrants (with similar characteristics) that do not do so.

Data
This study employs the Vietnam National Internal Migration (VNIM) dataset, which is a product of the Vietnam General Statistics Office. The dataset comprises two survey periods conducted in 2004 and 2015. Both migrant and non-migrants participated in the surveys. The surveys' samples were drawn randomly in 20 provinces in Vietnam using the stratified sampling technique to ensure that an adequate number of migrants and non-Regional informal institutions, local governance and internal migration in Vietnam migrants was included for each province. Figure 2 maps the sampled provinces. Appendix B in the supplemental data online provides the according list. It can be seen that the sampled provinces are distributed evenly across the north and south regions. Migrants are defined in the surveys as individuals who moved from one region (the minimum unit here being a district) to another within the last five years and satisfied at least one of the following conditions: (1) they have been living in the current region for more than a month; (2) they have been living in the current region for less than a month but intend to stay more than a month; or (3) they have been living in the current region for less than a month but within the last year they moved to another region for more than a month to work.
We acknowledge that internal migrants could be classified into two types: permanent (long term) and temporary (short term) (Liu & Xu, 2017). Since our study is concerned with job-seeking and business opportunity-seeking migration, we exclude migrants who have moved to their new locations in less than one year (i.e., potentially short-term migrants). Also, it takes time for entrepreneurs to yield earnings from their start-ups. As such, excluding short-term migrants and migrants whose relocating time is not sufficiently long to achieve regular earnings could reduce biases in estimation.
In 2004, the survey covered 5009 non-migrants and 4998 migrants; the numbers for 2015 were 3000 nonmigrants and 4969 migrants. The total observations were thus 17,976. However, after removing short-term migrants, and migrants who move not for job-seeking or business opportunity-seeking purposes (e.g., for-family and for-education), the number of observations reduces to 12,073, in which the 2004 subsample has 7801 observations and the 2015 subsample has 4272. The questionnaires contain a broad set of questions regarding individuals' socio-demographic backgrounds, current living and working conditions, and health conditions. For migrants, there was an additional section asking for their migration purposes, decision-making process, difficulties they faced in their new locales and their remittances.
To analyse the impacts of informal institutions (the north versus the south of Vietnam) on migration, we employ both the 2004 and 2015 surveys. However, when it comes to examining the effects of local governance on migration, we use only the 2015 survey. This is to match with the availability of the local governance datasetthe provincial governance and public administration performance index (PAPI). This index is a joint product of the Vietnam Centre for Community Support Development Studies (CECODES) and the United Nations Development Programme (UNDP). This dataset is a panel of provincial governance quality. The quality is scored from 0 to 10, with higher scores reflecting better governance quality. The calculation of the PAPI index is based on an annual survey of more than 14,000 citizens across Vietnam's provinces. In 2011 the PAPI index became available for all provinces and it has since been updated annually. The data are collected by fieldtrip surveys and face-to-face interviews and cover eight governance dimensions: (1) participation at local levels; (2) transparency; (3) vertical accountability; (4v) control of corruption; (5) public administrative procedures; (6) public service delivery; (7) environmental governance (from 2018); and (8) e-governance (also from 2018). 2 One may argue that migrants of different types are responsive to dissimilar dimensions of local governance. We will examine PAPI sub-indices in a robustness test. In the main analysis, following the argument of  that local governance settings are highly correlated and may influence each other, we use the weighted PAPI index as a general measure of local governance quality. It is noteworthy that we use the 2015 PAPI dataset to match with the time of the migration survey. As such, there are only six governance dimensions in this study. Appendix C in the supplemental data online provides definitions and summary statistics of the six governance dimensions.

Methods
To test the proposed hypotheses, we examine two models.

Model 1: Institutions and migration
Purposes of migration ig = b 0 + b 1 (South ig ) + b 2 (Local governance quality ig ) The dependent variable is Purposes of migration of individual i in province g, which is a categorical variable that takes value 0 for non-migrants, value 1 if migration for jobs and value 2 if migration for business opportunities. The first independent variable of interest is South, which takes value 1 for individuals living in the south of Vietnam and value 0 for individuals living in the north of Vietnam. The south versus north is identified using the 17th parallel. The second independent variable of interest is Local governance quality, which takes values of the PAPI index. The model also controls for a set of individual characteristics presented in Table 1. These characteristics, such as gender, age, marital status, education, ethnicity, desire to increase incomes, savings, debts and level of engagement in local activities, are likely to influence an individual's migration purposes (Liu, 2019;Li & Zahniser, 2002;Haughton et al., 2018;. Following Stark and Taylor (1991), we understand that migration decision is a function of individuals' initial relative deprivation. As such, we include relative incomes and its squared term to measure the gaps between individuals' earnings and the provincial average earnings as a proxy of relative deprivation. Also, the model controls for a year dummy v t and a set of provincial dummies v g to capture all time-invariant determinants at the macro-socio-economic and regional levels, respectively. It is noteworthy that the term v t is only available for the specifications estimating informal institutions since they include both the 2004 and 2015 samples. For specifications examining local governance (PAPI index), the term v t is not available since it includes only the 2015 survey. The m i is the error term. We estimate model 1 using the multinominal logit technique given the categorical nature of the dependent variable.

Model 2: Migration and earnings
Earnings ig = b 0 + b 1 (Migration across the north − South regions ig ) + b 2 (Difference in local goverance quality ig ) The dependent variable is the earnings of an individual i in province g, which is measured as the log of monthly earnings (millions of VND) in 2010 price. The independent variable Migration across the north-south regions is a dummy variable, which takes value 0 if the migrant moved within the north or south regions of Vietnam (the benchmark) and value 1 if the migrant moved across the north-south region. Meanwhile, the Difference in local governance quality variable indicates the difference in the PAPI scores (a proxy of local governance quality) of the migrant's place of origin and place of relocation (PAPI score of the destination minus PAPI score of the origin).
The term Individual characteristics is a vector of a subset of the covariates employed in model 1. We also include three additional variables: Jobs, which takes value 1 if the migrant is an employee and value 0 if he/she is an entrepreneur; rural to urban migration, which takes value 1 if a migrant moved from rural areas to urban areas, and 0 otherwise; and out of home province, which takes value 1 if migrants move out of their local province and value 0 if migrants move within their local province. These additional controls address how the types of jobs, scales (long versus short distance) and directions of migration may influence earnings (Liu & Xu, 2017;Nguyen et al., 2017a).
Moreover, we understand that a regime switch between the north/south may devaluate region-specific human capital, leading to reduced earnings. Ralston et al. (1999) show that managers in the north and south of Vietnam, even though having identical educational levels, possess some dissimilar regional-specific skills and knowledge. Therefore, we include unmatched skills and knowledge variable, which take value 1 if a migrant indicates that their skills and knowledge cannot match with their current jobs, 0 otherwise. Finally, the model controls for a year dummy v t , and a set of provincial dummies v g . The m i is the error term. We estimate model 2 using a robust ordinary least squares (OLS) regression.

Model 1: Institutions and migration
Regression results of model 1 are presented in Tables 2 and 3. The group of non-migrants is the benchmark. The variance inflation factor (VIF) statistic, which is 1.52, indicates no serious multicollinearity-related issues in our specifications. We report the odds ratios for the sake of interpretation. Table 2 shows the impacts of informal institutions. It can be seen that migrants moving for job-seeking purposes are less likely to come to the south (the odds ratios associated with South variable are less than 1 in odd-numbered columns), implying that job-seeking migrants are more likely to go to the north. Meanwhile, migrants are more likely to move into the south for business opportunityseeking purposes (the odds ratios associated with South variable are greater than 1 in even-numbered columns). It is noteworthy that even though the number of nonmigrant accounts for 66.3% of the total sample, after controlling for individual characteristics, the findings suggest that compared with non-migrants, migrants in the south are more likely to be entrepreneurs; meanwhile, compared with non-migrants, migrants in the north are more likely to be employees. The pattern is surprisingly consistent in both the 2004 and 2015 subsamples, given that there are 12 years of economic transformation and liberalization separating the two study periods. As such, Hypothesis 1 is supported. These findings reveal the differences in the migration types between regions having dissimilar norms and values. Since these informal institutional forces have deep social roots and are regionally sticky (Fritsch & Storey, 2014;Fritsch & Wyrwich, 2014), they are expected to exert consistent effects on migration decisions over time.
Turning to local governance, regression results in Table 3 indicate that provinces having a better governance quality attract more migrants (the odds ratio associated with local governance quality variable is statistically significantly greater than 1 in all columns). As such, Hypothesis 2 is supported. This positive effect could be explained by the fact that these provinces can offer migrants safer and better working/living conditions through a set of inclusive policies and effective execution of relevant regulations. Doan et al. (2018) further suggest that better provincial governance creates a healthier business environment, which draws more firms to the local markets. This, in turns, creates a greater demand for labour, which will attract migrants.
In terms of the control variables (Table 2), females are more likely than males to migrate for both job-seeking and business opportunity-seeking purposes. Younger people are more likely than older people to migrate. Also, people who are divorced or married are less likely to migrate than single people. People with higher education are less likely to migrate. People who want to increase their incomes are more likely to migrate for entrepreneurial opportunities; however, this tendency became less significant in 2015. People who have savings and are more embedded in their local communities are less likely to move. Finally, being consistent with Stark and Lucas (1988), we find that people are more likely to migrate when their relative incomes increase slightly (so that they have sufficient resources to move) but are less likely to do so when their incomes are extremely high.

Model 2: Migration and earnings
The regression results of model 2 are presented in Table 4. The VIF statistic, which is 1.85, indicates no serious multicollinearity-related issues in our specifications.
In Table 4, columns 1-3, the group of migrants moving within either the north or the south of Vietnam is the benchmark. The coefficients associated with migration across the north-south regions are negative and statistically significant in all specifications. This indicates that migrants moving across the north-south regions of Vietnam suffer from persistent earning reduction. To be specific, migrants who move across regions having different informal institutions earn less than migrants moving within regions that have similar informal institutions. As such, Hypothesis 3 is supported.
Turning to local governance, the regression result in Table 4, column 4, shows that migrants moving from provinces with a lower governance quality to provinces with a higher governance quality enjoy higher earnings (in comparison with earnings of migrants who do not move to provinces that have a higher governance quality). This is evidenced by the positive and statistically significant coefficient associated with the variable Difference in local governance quality between the origin and the destination. This positive effect could be attributed to the lower transaction costs in local markets , thereby improving firm performance and, subsequently, employees' incomes. Also, a well-structured and efficient set of local governance arrangements reduces red tape and corruption (Malesky et al., 2015), securing self-employed businesses and improving the survival and development of venturing activities. As such, migrants, regardless of whether they are employees or self-employed, may find more opportunities to boost their incomes in provinces that have better local governance quality. Hypothesis 4 is thus supported.

Robustness checks
We conduct a set of robustness tests and extensions, which are presented in Appendices D-F in the supplemental data online. First, we employ a multi-modelling approach to test whether or not our findings are sensitive to model specifications. The results confirm that findings are robust in specifications having different combinations of control variables.
Second, we use a principal component analysis (PCA) to examine whether or not job-seeking migrants and business opportunity-seeking migrants are sensitive to different local governance dimensions. It is found that job-seeking migrants are more responsive to participation at local levels, transparency, and administration procedures; meanwhile, business opportunity-seeking migrants are more sensitive to accountability, corruption, and public services.
Third, we investigate the subsample of Hanoi (the capital) and Hochiminh (the largest economic hub). The results show that the effect of informal institutions is weaker than the effect of local governance quality in these two cities. An explanation could be that these two metropolitan cities attract a high level of migrants of different origins (Haughton et al., 2018); as such, local indigenous values are diluted and mixed, leading to a convergence in values. Note: The estimator is multinomial logit, reporting the odds ratios. The benchmark group is non-migrants. Standard errors and test statistics are asymptotically robust to heteroskedasticity and are clustered to the provincial level. ***Significant at 1%; **significant at 5%; and *significant at 10%.

Regional informal institutions, local governance and internal migration in Vietnam 1201
Fourth, since migrants moving not only for jobs and business opportunities but also for family and education, we also explore these samples. It is found that the north attracts more for-family and for-education migrants than the south.
Fifth, a dummy variable indicating informal institutions associated with the north and south may insufficiently strong. As such, we employ the World Value Survey data to directly measure collectivism and proentrepreneurship values associated with the two regions and use them to retest the hypotheses. The findings are consistent with those presented in the main analysis. 3 Finally, we also test the moderating effect of informal institutions on the relationship between local governance and migration. The odds ratios associated with the interaction terms are insignificant, indicating that regions with strong local governance attract migrants regardless of their locations being north or south.

CONCLUSIONS
This study examines the importance of institutional forces, namely regional institutions and local governance, to internal migration. It investigates the impact of these forces on (1) a set of migration purposes (job-seeking and business opportunity-seeking), and (2) migrants' earnings. The context of analysis is Vietnaman emerging country with substantial urbanization and industrialization but heterogeneous institutional settings across its regions.
In terms of informal institutions, we proposed that the differences in regional norms, values, and beliefs (which are caused by a historical trajectory) between the north and south of Vietnam may trigger dissimilar types of migration into each region. We also expected that migration across these two regions would lead to discrimination and increased adaptation costs, thereby reducing earnings for migrants. In terms of local governance, we proposed that governance quality is positively associated with in-migration and that migrants moving from regions with a lower quality of governance to regions with a higher quality of governance earn more.
We examined these questions using a unique set of internal migration data in Vietnam in 2004 and 2015.
Our findings point to the importance of the surrounding institutional environments in shaping migration purposes. Specifically, we found that in the north of Vietnam, where collectivism and socialism are the dominant values, migrants are more likely to move in the region for job-seeking purposes. Meanwhile, in the south of Vietnam, where individualism and pro-entrepreneurship are the key values, migrants are more likely to move in for business opportunity-seeking purposes. This pattern is consistent over time (from 2004 to 2015), despite Vietnam's remarkable economic transformation and liberalization over that period. More importantly, we find that migration across the two regions does not yield financial benefits. This finding is consistent with a research strand in regional economics highlighting the effects of north/south informal institutions on the socio-economic performance of the two regions Dell et al., 2018).
Our findings also suggest that local governance quality plays an essential role in attracting migrants. More importantly, migrants moving from regions with a lower governance quality to regions with a higher governance quality can increase their earnings. However, this positive effect is less obvious on job-seeking migrants than it is on business opportunity-seeking migrants.
This study provides some policy implications for developing countries. First, when designing migration regulations, policymakers should take into account the influences of local and regional informal institutions that shape migrants' moving purposes. By doing this, governments can improve the efficiency of labour force Note: The estimator is multinomial logit, reporting the odds ratios. The benchmark group is non-migrants. Standard errors and test statistics are asymptotically robust to heteroskedasticity and are clustered to the provincial level. ***Significant at 1%; **significant at 5%; and *significant at 10%.
allocations, which subsequently helps migrants as well since different types of migration are associated with specific sets of difficulties and need specific aids from the governments (Liu & Xu, 2017). Second, the discrimination/adaption costs experienced by migrants who move across regions within a country harms the effectiveness of the labour force allocation; when labours can freely move across regions without suffering from informal Note: The number of observations includes migrants only. The benchmark group in columns 1-3 is within-north or within-south migration. The estimator is robust ordinary least squares (OLS). Standard errors and test statistics are asymptotically robust to heteroskedasticity and are clustered to the provincial level. ***Significant at 1%; **significant at 5%; and *significant at 10%.

Regional informal institutions, local governance and internal migration in Vietnam
obstacles, it is beneficial to both the sending and receiving regions (De Haas, 2010;Zhang, 2019). Hence, authorities should intervene by promoting values such as diversity within local communities. This can be done by implementing subsidy schemes and policies that encourage local businesses to employ non-local employees. This study is not without limitations that should be acknowledged but which also provide potential avenues for future research. First, the dataset employed in this study is country-specific. One of the main weaknesses of a country-specific research design is that we only observe within-country effects, which may be influenced by social embeddedness. Future research should therefore retest the validity of our findings using a multi-country dataset with longer survey periods. In addition, due to data limitations, we are restricted to two sets of study periods (2004 and 2015). Finally, the investigations of earnings in this study are pure cross-sectional, in which migrants' earnings are compared with those of non-migrants. It would be more relevant to compare earnings of migrants between the pre-and post-migration periods. Due to the cross-sectional nature of the dataset, we are unable to conduct a two-period investigation of migrants' earnings. Future research should design a longitudinal study to examine this issue deliberately.