Applied Economics Letters

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/rael20

Relationship between household income and

socio-political capital in rural Vietnam: a panel
quantile regression approach

Myeong Jun Kim, Tram T. H. Nguyen & Sung Y. Park

To cite this article: Myeong Jun Kim, Tram T. H. Nguyen & Sung Y. Park (2021): Relationship
between household income and socio-political capital in rural Vietnam: a panel quantile regression
approach, Applied Economics Letters, DOI: 10.1080/13504851.2021.1898533

To link to this article: https://doi.org/10.1080/13504851.2021.1898533

Published online: 09 Mar 2021.

Submit your article to this journal

Article views: 50

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=rael20
APPLIED ECONOMICS LETTERS
https://doi.org/10.1080/13504851.2021.1898533

ARTICLE

Relationship between household income and socio-political capital in rural

Vietnam: a panel quantile regression approach
Myeong Jun Kima, Tram T. H. Nguyenb and Sung Y. Parkb
a
School of Economics and Finance, South China University of Technology, Guangzhou University Town, Guangzhou Shi, Guangdong Sheng,
China; bSchool of Economics, Chung-Ang University, Seoul, South Korea

ABSTRACT KEYWORDS
This study uses the Vietnam Access to Resources Household Survey (VARHS) data from 2008 to Social capital; political
2014 to analyse the relationship between socio-political capital and household income in rural capital; household income;
Vietnam. We consider six types of socio-political capital (connections with others that create access VARHS; Vietnam; panel
to useful information and/or benefits) namely Communist Party membership, membership in mass quantile regression
organizations (MOs) and voluntary groups, connections with officials, financial assistance, and JEL CLASSIFICATION
trust. Communist Party membership and financial assistance have significant positive effects on D73; H7; O12; O53; P16
household income in the pooling, random, and fixed effects models and at all quantile levels. The
effect of trust is also positive and significant in all models but only significant at high quantile
levels. Whereas connections with officials show significantly positive effects in the pooling and
random effects models and at most quantile levels, the effects of MOs and voluntary groups are
negative and significant for some models.

1 Introduction
Socio-political capital plays an important role in political capital, strengthened land property rights
determining household income, especially, in and access to credit and transfers. Additionally,
developing countries because it may help indivi­ politically connected households have easier access
dual households in obtaining access to good jobs or to public benefits than other households (Newman
to cheaper supplies of credit and labour and/or and Zhang 2015). Vo (2018) also used local group
sharing information, thereby increasing their abil­ memberships in the Communist Party, Women’s
ity to invest and to profit from their businesses. Union, and Farmers’ Union to measure house­
Coleman (1988) identifies three types of socio- holds’ social capital and found that participation
political capital that may be resources for improv­ in these organizations reinforces the Vietnamese
ing individual performance: information sharing, households’ ability to overcome negative shocks
obligations and trust, and social norms. and is associated with a higher level of saving.
Some empirical studies define social capital All these previous studies’ findings indicate that
based on the context of the study and the avail­ social and political capital is quite closely related to
ability of data. Newman, Tarp, and Van Den income-related variables. However, they also show
Broeck (2014) define social capital as information the linear relationship between these variables in
sharing and the elimination of information asym­ a conditional mean framework. Clearly, the effects
metries through membership in the Women’s of social and political capitals differ with different
Union in the case of Vietnam. Membership in levels of incomes. For example, Yu and Wang
high-quality networks, such as the Women’s (2018) indicate that the social networks are more
Union leads to higher levels of saving in formal important for the high-income class and less
financial institutions and productive investments, important for the low-income class on income
thereby positively affecting household income. increases.
Markussen and Tarp (2014) found that personal Theoretical and empirical studies of socio-
connections to local government officials, a form of political capital have identified important

CONTACT Sung Y. Park [email protected] School of Economics, Chung-Ang University, 84 Heukseok-Ro, Dongjak-Gu, Seoul, South Korea
© 2021 Informa UK Limited, trading as Taylor & Francis Group
2 M. J. KIM ET AL.

contributions to household well-being as well discusses the empirical results. Section 4 concludes
as economic development. In this study, we the paper.
examined the effects of social capital on house­
hold income in rural Vietnam by using the
Vietnam Access to Resources Household I. The empirical model
Survey (VARHS) data from 2008 to 2014
Consider the following linear regression model
along with some variables identified in many
with panel data,
previous studies: Communist Party member­
ship, membership of mass organizations
yij ¼ x0 it β þ αi þ uit ; i ¼ 1; . . . n;
(MOs: Women’s Union, Farmers’ Union, (1)
t ¼ 1; . . . ; T:
Youth Union, and Veterans’ Union), member­
ship of voluntary groups (business association,
where αi denotes unobserved heterogeneity for the
religious groups, groups for the elderly and
ith household and uit is the usual random distur­
many others), connections to officials, and
bance. The conditional quantile functions of yit
trust in unspecified ‘strangers’. We analysed
given xit can be represented by
this relationship using a panel quantile regres­
sion model as well as three linear regression
Qyit ðτjxit Þ ¼ αi þ x0 it βðτÞ; τ 2 ð0; 1Þ: (2)
models that provide a more complete picture
of the relationship. In the linear regression In model (2) the α’s have a pure location shift effect
model, Communist Party membership, connec­ on the conditional quantiles of the response. The
tion to officials, financial helpers, and trust effects of the covariates, xit , are permitted to
have a significant positive effect on household depend upon the quantile, τ, of interest, but the
income in the pooling, random, and fixed α’s do not. To estimate the model (2) for several
effects models, with the exception of officials quantiles simultaneously, Koenker (2004) pro­
in the fixed-effects model. Meanwhile, partici­ posed using a penalized fixed effects approach,
pating in mass organizations (MOs) and volun­
tary groups have a significant negative effect on q X
X n X
T
household income in the pooling and random min ωk ρτk ðyit αi x0 it βðτ k ÞÞ
effect models. In the panel quantile regression α;βðτÞ
k¼1 i¼1 t¼1
model, Communist Party membership and con­ Xn
þλ jαi j; (3)
nection to officials have significant positive i¼1
effects on household income at almost all quan­
tile levels. Financial helpers also have where ρτ ðuÞ ¼ uðτ Iðu < 0ÞÞ denotes Koenker
a significant positive effect on household and Bassett (1978) piecewise linear quantile loss
income at all quantile levels; whereas, the esti­ function. The weights ωk control the relative influ­
mated values of trust are only significant at ence of the q quantiles τ1 ; . . . ; τ q , on the estimation
a high quantile level. However, membership in P
of the αi parameters and ni¼1 jαi j is the ,1 penalty
MOs have significant negative effects on house­ that controls the variability introduced by the large
hold income at all quantile levels, which is the number of estimated parameters. λ is the tuning
opposite of what we expected to prove. By parameter to control the individual effect. If the λ
contrast, membership of voluntary groups term tends to zero, then the penalty term disap­
does not have any significant value at any pears and we obtain the usual fixed effects estima­
quantile level. tor. Conversely, if the λ term tends to infinity, then
The remainder of the paper is organized as fol­ we obtain an estimate of the model without indivi­
lows. In Section 2, we discuss the panel quantile dual effects. We set λ = 1 similar to Damette and
regression model. Section 3 describes the data and Delacote (2012) and Zhu et al. (2016).1

1
We considered λ from 0.1 to 0.9 and found that this set of λ did not affect the precision of the β estimates.
 APPLIED ECONOMICS LETTERS 3

The empirical model for the relationship obtaining privileges and specific information,
between social capital and household incomes is among others. Thus, Communist Party member­
as follows: ship and connections to officials are considered
linking. Second, based on the strength of ties and
ln yit ¼ z0 it β þ x0 it γ þ αi þ δt þ uit ; (4)
the social distance between the household and their
where yit is the real per capita income of household network member, financial helpers belong to bond­
i in year t, zit is a vector of social capital variables ing (strong ties to people of similar social stand­
including Communist party membership, mem­ ing). Third, measures of ‘generalized trust’ are
bership of mass organizations and other voluntary commonly used as a measure of bridging. Fourth,
groups, connections to officials (having we define that membership of MOs and voluntary
a household member, relative or friend who is groups may belong to both linking and bridging
a local government official), financial helpers social capital. For example, as Markussen and Tarp
(number of individuals willing to lend money in (2014) mentioned, households in rural Vietnam
case of emergency), and score on the trust index. xit that participate in the Women’s or Farmers’
is a vector of variables including years of school Unions are easily able to access sponsored loans
and age of household head, female household head, from banks and other community information.
and Kinh (household head belonging to the ethnic However, because membership of voluntary
majority in Vietnam). αi is a household fixed effect groups is less common than membership of MOs,
and δt is a dummy variable for the year. uit is it may be considered ‘weak’ linking and bridging.
a disturbance term. We consider four types of social ties with six socio-
political capital measures: i) linking (Communist
Party membership and connections to officials), ii)
II. Data and empirical results bridging (trust), iii) bonding (financial helpers),
The Vietnam Access to Resources Household and iv) linking and bridging (membership of
Survey (VARHS) data from 2008 to 2014 are used mass organizations (MOs) and voluntary groups).
to study the relationship between social capital and Table 1 shows the comparison of the sample
household income in rural Vietnam.2 Markussen means calculated under different values of six social
(2015) classified these variables broadly into three capital variables. We have a different sample ratio
categories: linking (ties to people in power), bond­ for each variable considered; for example, we can
ing (within-group ties), and bridging (between- divide the sample into two groups: those that are
group ties). First, if someone is a member of the Communist Party members and those that are not.
Communist Party or knows a public official, it is The sample consists of 9% Communist Party mem­
possible to increase household income by easily bers and 91% Non-Communist Party members.

Table 1. Comparison of sample means.

Party member MOs Voluntary groups Official Financial helper Trust All
A=0 A=1 A=0 A � 1 A=0 A � 1 A=0 A=1 A=0 A � 1 A=0 A=1
Log income 9.559 10.072 9.569 9.619 9.592 9.641 9.537 9.758 9.616 9.606 9.587 9.630 9.607
Party member 0.000 1.000 0.075 0.098 0.084 0.114 0.058 0.167 0.115 0.091 0.093 0.092 0.093
MOs 1.296 1.370 0.000 1.737 1.424 1.013 1.224 1.472 1.007 1.326 1.301 1.305 1.303
Voluntary groups 0.310 0.409 0.473 0.268 0.000 1.078 0.301 0.357 0.304 0.320 0.327 0.309 0.319
Official 0.291 0.572 0.243 0.341 0.303 0.348 0.000 1.000 0.172 0.328 0.333 0.297 0.317
Financial helper 4.065 5.344 3.366 4.455 3.979 4.669 3.494 5.670 0.000 4.518 4.310 4.035 4.183
Trust 0.462 0.458 0.469 0.459 0.465 0.454 0.475 0.434 0.459 0.462 0.000 1.000 0.462
Years of schooling, hh head 6.463 9.004 5.198 7.198 6.907 6.203 6.328 7.498 6.351 6.726 6.767 6.618 6.699
Age of hh head 53.622 57.438 59.347 52.186 49.487 64.661 53.970 53.986 55.779 53.831 53.867 54.100 53.975
Female hh head 0.224 0.195 0.339 0.182 0.172 0.339 0.240 0.182 0.234 0.221 0.225 0.218 0.222
Kinh 0.801 0.798 0.819 0.795 0.771 0.872 0.801 0.800 0.784 0.802 0.816 0.783 0.801
Notes: The means are the weighted average using time-series sample weights that are derived by Vietnam Access to Resources Household Survey (VARHS) data
from 2008 to 2014. A denotes participation in a group, where A = 0 for those who do not participate, otherwise A = 1 (or � 0) and hh denotes household.

2
The survey was conducted every 2 years from 2006 to 2014 across 12 provinces in Vietnam. Because several questions changed between 2006 and 2008, we
use the data from 2008 to 2014 in terms of continuity. The objective and structure of the data set are fully explained on data web-page (https://www.wider.
unu.edu/database/viet-nam-data).
4 M. J. KIM ET AL.

The ratio of Communist Party members is much political capital helps to increase household
smaller than the other ratios, for example, MOs income. By contrast, membership of MOs have
(75%) vs. non-MOs (25%), voluntary groups a significant negative effect on household income
(30%) vs. non-voluntary groups (70%), official in the pooling and RE models. However, this effect
(32%) vs. non-official (68%), financial helper is insignificant in the FE model.
(46%) vs. non-financial helper (54%), and trust We also use a panel quantile regression model to
(46%) vs. non-trust (54%). Because Vietnam’s find the effect of these variables on household
local government has a huge effect on all areas of income in more detail. The estimation results of
rural economic life (Markussen and Tarp 2014), the quantile regression are shown in Table 3 for
becoming a Communist Party member may be nine τs, τ ¼ 0:1; . . . ; 0:9. Figure 1 also summarizes
more difficult than being a member of MOs or the estimated regression quantile over nine quan­
voluntary groups. Thus, we may expect that tile levels with 95% confidence bands.3
Communist Party membership has a more signifi­ Communist Party membership, corresponding
cant positive effect on household income than MOs to linking, has significant positive effects on house­
and voluntary groups do. hold income at all quantile levels. The marginal
Table 2 presents the panel data regression effect of Communist Party membership decreases
results. Three types of panel estimations are con­ gradually as the quantile level increases. This
sidered: (i) pooling, (ii) random effect estimation implies that joining the Communist Party has
(RE), and (iii) fixed effect estimation (FE). The a greater effect in lower-income households than
results indicate that Communist Party member­ for higher-income households. This is because
ship, connections to officials, financial helpers, members of high-income households are more
and trust have a significant positive effect. likely to participate in non-agricultural businesses
However, connection to officials does not have that generate higher incomes. Communist Party
a significant effect for the FE. Significant positive membership has relatively less effect on lower-
effects are quite predictable because socio- income households.
In addition, another measure of linking – con­
nections with government officials – has
Table 2. The estimated results: panel data regression. a significant positive effect on household income
Pooling RE FE over higher and lower quantile levels. These results
(Intercept) 7.506��� 7.231��� – are in the same vein as Markussen and Tarp (2014)
(0.163) (0.167) –
Party member 0.321��� 0.257��� 0.103��� who find that personal connections with local gov­
(0.036) (0.032) (0.038)
MOs −0.062��� −0.026��� −0.006
ernment officials strengthen land property rights
(0.011) (0.010) (0.011) and make access to credit and transfers easier. Our
Voluntary groups −0.002 0.010 −0.005
(0.022) (0.019) (0.020) results show that such effects are quite strong for
Official 0.118��� 0.075��� 0.028 τ ¼ 0:1 and τ ¼ 0:8
(0.020) (0.017) (0.018)
Financial helper 0.012��� 0.011��� 0.009��� Financial helpers, a bonding social capital, have
(0.002) (0.001) (0.001) a significant positive effect on household income at
Trust 0.076��� 0.080��� 0.030�
(0.017) (0.015) (0.016) all quantile levels. Moreover, such effects are rela­
Years of schooling, hh head 0.053��� 0.050��� 0.014��� tively constant across different quantile levels.
(0.004) (0.003) (0.005)
Age of hh head 0.045��� 0.054��� 0.024�� However, the estimated values of financial helpers
(0.006) (0.006) (0.010)
Age2 of hh head −0.038��� −0.043��� −0.027���
have the smallest significant positive effect on
(0.005) (0.005) (0.009) household income among all the socio-political
Female hh head 0.069�� 0.089��� 0.122��
(0.029) (0.028) (0.054) capital variables because informal loans from help­
Kinh 0.491��� 0.469��� 0.208� ers are usually used for consumption smoothing,
(0.032) (0.032) (0.125)
not entirely for production and asset accumulation
Notes: The heteroscedasticity and autocorrelation consistent (HAC) standard
errors are calculated and standard errors are in parentheses. ***, **, and *
denote the 1%, 5%, and 10% levels of significance, respectively.

3
To save space, we only report the estimated results of the socio-political capital variables. The plots of the estimated quantile models’ results are available in
the on-line version of the supplementary materials.
 APPLIED ECONOMICS LETTERS 5

Table 3. The estimated results: panel quantile fixed effect.

τ(0.1) τ(0.2) τ(0.3) τ(0.4) τ(0.5) τ(0.6) τ(0.7) τ(0.8) τ(0.9)
Party member 0.318��� 0.261��� 0.272��� 0.253��� 0.235��� 0.238��� 0.213��� 0.180��� 0.153���
(0.040) (0.039) (0.037) (0.033) (0.033) (0.033) (0.030) (0.034) (0.039)
MOs −0.043��� −0.027�� −0.025�� −0.021�� −0.022�� −0.025�� −0.033��� −0.035��� −0.047���
(0.014) (0.012) (0.011) (0.010) (0.011) (0.011) (0.011) (0.012) (0.013)
Voluntary groups 0.039 0.031 0.034 0.006 −0.004 0.005 −0.001 −0.028 −0.026
(0.028) (0.023) (0.023) (0.020) (0.020) (0.020) (0.018) (0.020) (0.027)
Official 0.061�� 0.058�� 0.034 0.037� 0.038�� 0.016 0.035� 0.062��� 0.049��
(0.028) (0.024) (0.023) (0.021) (0.019) (0.019) (0.020) (0.022) (0.024)
Financial helper 0.011��� 0.009��� 0.008��� 0.009��� 0.009��� 0.009��� 0.009��� 0.009��� 0.010���
(0.002) (0.002) (0.001) (0.002) (0.002) (0.001) (0.001) (0.002) (0.002)
Trust 0.032 0.011 −0.010 0.004 0.001 0.016 0.029 0.017 0.051��
(0.024) (0.021) (0.018) (0.019) (0.016) (0.015) (0.018) (0.020) (0.026)
Years of schooling, 0.040��� 0.045��� 0.045��� 0.043��� 0.043��� 0.040��� 0.042��� 0.040��� 0.041���
hh head (0.005) (0.004) (0.003) (0.003) (0.003) (0.003) (0.003) (0.003) (0.005)
Age of hh head 0.040��� 0.037��� 0.033��� 0.034��� 0.040��� 0.043��� 0.044��� 0.042��� 0.040���
(0.007) (0.007) (0.006) (0.006) (0.005) (0.005) (0.006) (0.006) (0.007)
Age2 of hh head −0.037��� −0.033��� −0.030��� −0.030��� −0.035��� −0.037��� −0.039��� −0.037��� −0.036���
(0.006) (0.006) (0.005) (0.005) (0.005) (0.004) (0.005) (0.005) (0.007)
Female hh head 0.033 0.069�� 0.065�� 0.055 0.075��� 0.073��� 0.063��� 0.045 0.059�
(0.033) (0.029) (0.027) (0.028) (0.025) (0.024) (0.027) (0.028) (0.034)
Kinh 0.534��� 0.490��� 0.507��� 0.527��� 0.513��� 0.511��� 0.501��� 0.493��� 0.487���
(0.037) (0.029) (0.029) (0.026) (0.024) (0.021) (0.023) (0.025) (0.035)
Notes: The Robust standard errors are calculated and standard errors are in parentheses. ***, **, and * represent the 1%, 5%, and 10% levels of significance,
respectively.

Figure 1. Panel quantile regression results. Notes: The solid lines denote the estimated panel quantile effect at each quantile level and
the dashed lines correspond to the 95% confidence bands. The standard errors are calculated by using robust covariance matrix
estimates.

as formal loans from credit institutions are households are estimated to earn a higher income
(Barslund and Tarp 2008). than other households, and conversely, higher-
Although the effects of trust (commonly consid­ income or wealthy households are more likely to
ered bridging social capital) are significantly posi­ be more trusting than poorer households.
tive for the pooling, random, and fixed effects By contrast, membership in MOs has significant
models, the panel quantile regression results show negative effects on household income at all quantile
that the estimated marginal effect is only significant levels, whereas the effect of voluntary groups is
for very high quantile levels, τð0:9Þ. This implies insignificant at all quantile levels. Here, we should
that households may benefit from being trusting mention that the estimated results do not exclude
because trust includes engaging in profitable but that group members affect other aspects of house­
risky transactions (Markussen 2015). High-trust hold well-being, or that there is a positive social
6 M. J. KIM ET AL.

return to group activities, as Markussen (2015) personal or social positive effects based on social
mentioned. There is no doubt that membership in capital can also be adversely affected by social capi­
these groups assists households to increase their tal. Thus, it would be advantageous to consider both
ability to access information and formal and infor­ simultaneously, but this is difficult to do because of
mal loans; however, Banerjee et al. (2018) argued data characteristic limitations. In this study, we
that by charging a reduced rate below the inflation focused on the relationships affected by the former.
level, the preferential credit programmedis­ To estimate the relationship between socio-political
courages savings and encourages low return invest­ capital and household income, we use the Vietnam
ments. Consequently, preferential credit does not Access to Resources Household Survey (VARHS)
contribute to increasing household income. data from 2008 to 2014.
Furthermore, access to credit does not mean that The estimated results show that linking
households are not credit-constrained (Dinh, (Communist Party membership and connections
Dufhues, and Buchenrieder 2012). Indeed, the with officials) and bonding (financial helpers)
average loan size of Vietnam Bank for Social have significant positive effects on household
Policies (VBSP) for these households was VND income in the pooling, random, and fixed effects
3,175 million (about USD 182 at the time of the models, and at all quantile levels (except for con­
survey), which is about 1.5 times the per capita nections with officials in the fixed-effects model).
poverty line (Nguyen et al. 2007), is too small, The effects of trust (defined as a kind of bridging
and may not be sufficient to cover farmers’ needs. social capital) are also significantly positive for the
Coleman (2002) attributes the negative effect to the pooling, random, and fixed effects models;
loans being too small for investment, the loans whereas the panel quantile regression results
being used for consumption, and households turn­ show that the estimated marginal effect is only
ing to moneylenders to finance the repayments, significant at very high quantile levels. Among
leading to a vicious circle. these variables, the significant positive estimated
In summary, the results are consistent with values increase in the following order: financial
our expectation of positive bonding (financial helpers, trust, connections with officials, and
helpers) and linking (Communist Party mem­ Communist Party membership. That is, among
bership and connections with officials). the three broad categories, linking had the greatest
Especially, the estimated value was largest for positive influence on household income. By con­
Communist Party membership and smallest for trast, the effects of MOs and voluntary groups
financial helpers, on average. In terms of brid­ (measures of linking and bridging social capital)
ging, trust shows positive effects on household are negative and significant for some models. The
income, only for households at a very high possible explanations for these negative results are
quantile level. Lastly, for the linking and brid­ the benefits gained from those linking and brid­
ging combination, the estimated values of MOs ging variables, such as preferential credit program
are negative and significant and those of volun­ accessibility, may not contribute to increasing
tary groups are not significant. Although this is household income (Banerjee et al. 2018;
opposite to our original expectations, it can be Coleman 2002).
explained by the ineffectiveness of preferential
loans usage that households accessed via those
organizations. Disclosure statement
No potential conflict of interest was reported by the authors.
III. Concluding remarks
Some socio-political capital has a positive effect on References
the social development of a country or region. Here,
Banerjee, A., E. Duflo, R. Glennerster, and C. Kinnan. 2018. “The
positive effect is used in a broad sense, and can refer Miracle of Microfinance? Evidence from a Randomized
to an increase in household income, or an increase Evaluation.” American Economic Journal: Applied Economics 7
in social safety and infrastructure. Conversely, (1): 22–53.
 APPLIED ECONOMICS LETTERS 7

Barslund, M., and F. Tarp. 2008. “Formal and Informal Rural Markussen, T., and F. Tarp. 2014. “Political Connections and
Credit in Four Provinces of Vietnam.” Journal of Development Land-related Investment in Rural Vietnam.” Journal of
Studies 44 (4): 485–503. doi:10.1080/00220380801980798. Development Economics 110: 291–302. doi:10.1016/j.
Coleman, B. E. 2002. “Microfinance in the Northeast jdeveco.2014.01.011.
Thailand: Who Benefits and How Much?” ERD Working Newman, C., F. Tarp, and K. Van Den Broeck. 2014. “Social
Paper, Asian Development Bank, Manila, Philippines. Capital, Network Effects, and Savings in Rural Vietnam.”
Coleman, J. S. 1988. “Social Capital in the Creation of Human Review of Income and Wealth 60 (1): 79–99. doi:10.1111/
Capital.” American Journal of Sociology 94: S95–S120. roiw.12061.
doi:10.1086/228943. Newman, C., and M. Zhang. 2015. “Connections and the
Damette, O., and P. Delacote. 2012. “On the Economic Allocation of Public Benefits.” Working Paper, UNU-
Factors of Deforestation: What Can We Learn from WIDER, Helsinki, Finland.
Quantile Analysis?” Economic Modelling 29 (6): Nguyen, C., D. Bigman, M. V. Den Berg, and T. Vu. 2007. “Impact
2427–2434. doi:10.1016/j.econmod.2012.06.015. of Micro-credit on Poverty and Inequality: The Case of the
Dinh, Q. H., T. B. Dufhues, and G. Buchenrieder. 2012. “Do Vietnam Bank for Social Policies.” MPRA Paper, 54154.
Connections Matter? Individual Social Capital and Credit Vo, T. T. 2018. “Social Capital and Household Vulnerability:
Constraints in Vietnam.” European Journal of Development New Evidence from Rural Vietnam.” Working Paper,
Research 24 (3): 337–358. doi:10.1057/ejdr.2012.11. UNU-WIDER, Helsinki, Finland.
Koenker, R. 2004. “Quantile Regression for Longitudinal Yu, X., and X. Wang. 2018. “Will Social Networks Become
Data.” Journal of Multivariate Analysis 91 (1): 74–89. Less Important.” Applied Economics Letters 25 (16):
doi:10.1016/j.jmva.2004.05.006. 1117–1120. doi:10.1080/13504851.2017.1400643.
Koenker, R., and G. Bassett. 1978. “Regression Quantiles.” Zhu, H., L. Duan, Y. Guo, and K. Yu. 2016. “The Effects of
Econometrica 46 (1): 33–50. doi:10.2307/1913643. FDI, Economic Growth and Energy Consumption on
Markussen, T. 2015. “Social and Political Capital in Rural Carbon Emissions in ASEAN-5: Evidence from Panel
Vietnam.” Technical Report 87. United Nations University Quantile Regression.” Economic Modelling 58: 237–248.
World Institute for Development Economics Research. doi:10.1016/j.econmod.2016.05.003.