DOI: 10.1111/ijsw.
12318
Int J Soc Welfare 2018: 28: 63–76
Examining the effects of the Durunuri
programme on low-wage workers’ social
insurance coverage in South Korea
Park M., Cho R.M. Examining the effects of the Durunuri
programme on low-wage workers’ social insurance coverage
in South Korea
This study examined the effects of the Durunuri programme
in Korea, a Making Work Pay policy initiated in 2012 that
subsidises social insurance contributions of low-wage
workers in workplaces with fewer than 10 employees. Using
a nationally representative sample of Korean low-wage
workers between the years 2010 and 2015, the study employs
a difference-in-differences method to isolate the effect of the
Durunuri programme on workers’ subscription rate to the
national pension and employment insurance programmes.
Results indicated that, compared with ineligible low-wage
workers in workplaces with 10 or more but fewer than 30
employees, eligible low-wage workers in workplaces with
fewer than 10 employees were found to experience an
increase in the odds of being subscribed to both the national
pension and employment insurance programmes during the
first, second and third year following implementation of the
Durunuri programme. Policy implications as well as future
policy modifications are further discussed.
Previous research has emphasised the role of govern-
ment in resolving social problems such as unemploy-
ment and poverty; thus, many countries have
attempted to solve such problems through various
governmental mechanisms (Marx, Vanhille, &
Verbist, 2012). Public assistance programmes are one
type of such policies aiming to alleviate poverty
through the provision of financial support. However,
provision of generous assistance when out of work
has been criticised for reducing work incentives and
thus increasing welfare dependency.
As an alternative, ‘Making work pay’ (MWP) poli-
cies have been introduced not only to reduce poverty
but also to boost work incentives and decrease unem-
ployment rates. Specifically, various types of MWP
policies have been advocated for, and implemented to
address, these issues in OECD countries in recent
years (Bargain, Caliendo, Haan, & Orsini, 2010;
Dagsvik, Jia, Orsini, & Van Camp, 2011; Van der
Linden, 2016).
MWP policies are generally targeted at people
who face the highest unemployment risks (Bargain
et al., 2010). Typical MWP policies are workfare
programmes that provide work incentives and main-
tain certain levels of social protection. The goal is
Int J Soc Welfare 2018: 28: 63–76
© 2018 The Authors. International Journal of Social Welfare © 2018 Akademikerförbundet SSR (ASSR) and John Wiley & Sons Ltd
INTERNATIONAL
JOURNAL OF
SOCIAL WELFARE
ISSN 1369-6866
Mikyung Park, Rosa Minhyo Cho
Graduate School of Governance, Sungkyunkwan University,
Seoul, Republic of Korea
Key words: international social work, working poor, making
work pay, Durunuri programme, social insurance, difference-
in-differences, South Korea
Rosa Minhyo Cho, Department of Public Administration,
Graduate School of Governance, Sungkyunkwan University,
25-2, Sungkyunkwan-Ro, Jongno-Gu, Seoul, Republic of
Korea
E-mail: chomh@skku.edu; qkralrud14@skku.edu
Accepted for publication 26 February 2018
to boost the incomes of poor families while simulta-
neously enhancing the incentive to work. As a
result, MWP policy instruments generally take on
the form of tax credits on low earnings as well as
subsidies on social security contribution. Examples
of the former include the earned income tax credit
(EITC) in the USA, Working Families’ Tax Credit
in the UK, while examples of the latter include
‘mini-jobs’ in Germany and the Work Bonus in
Belgium.
The aim of the present study was to evaluate a
recent MWP policy in South Korea (hereinafter,
Korea) called the Durunuri programme, which subsi-
dises the social security contribution fees of low-
wage employees working in small businesses. The
Durunuri programme was the second MWP policy
implemented in Korea in July 2012, followed by the
EITC in 2009. Although MWP policies are typically
introduced in the Western context to address concerns
related to low employment of the low-skilled (Van
der Linden, 2016), the purpose of the Durunuri pro-
gramme was slightly different. That is, issues of
unemployment were less of a concern regarding the
Korean low-skilled population where employment
rates of those with less than a high school diploma
63
1
Park & Cho
are generally higher than their college graduate coun-
terparts.1 Rather, a key social issue over the past cou-
ple of decades regarding the low-skilled has been the
lack of social protection provided to this group chal-
lenging the provision of a minimum level of welfare
for all. To address such limitations, the primary pol-
icy goal of the Durunuri programme was to extend
coverage of social insurance programmes to the low-
skilled population who were particularly at risk of
being excluded by reducing both employers’ and
employees’ social contribution fees (Yoo, Kang, &
Choi, 2016). Other secondary goals of the pro-
gramme included promoting the transition of informal
labour to the formal sector, increasing the employ-
ment rate of low-wage workers and redistributing
income, thereby improving the well-being of low-
wage workers and their employers in small busi-
nesses (Yoo, Kang, & Choi, 2013).
As mentioned above, low coverage of social insur-
ance programmes in Korea is not a problem uniform
across all groups of workers (Koh, 2015). Rather,
employees with non-standard employment and those
in small firms have been found to be particularly vul-
nerable to such risks (Cho, Kim, & Kwon, 2008;
Koh, 2015). Scholars have long criticised the govern-
ment for tolerating such employer practices through
weak enforcement of laws as well as through exclud-
ing non-standard workers from social insurance-
related regulations (Cho et al., 2008; Seo & Back,
2014). Yet, minimum access to social protection
among the disadvantaged is a phenomenon common
in societies that are characterised by welfare ideolo-
gies such as ‘welfare developmentalism’ or ‘producti-
vist welfare capitalism’. These terms have been
frequently used to describe the Korean welfare state,
especially during the periods of economic develop-
ment in the 1980s and 1990s (Holliday, 2000; Kwon
& Holliday, 2007). In productivist welfare regimes,
social policy is essentially a means for economic
growth and is subordinate to economic objectives
(Holliday, 2000). Naturally, such societies develop
social policies that prioritise the productive sectors
and use social investment strategies to strengthen
human capital development, often resulting in the
neglect of its most disadvantaged population.
1
Information on unemployment rates by educational attain-
ment levels from 2000 to 2016 is available on the Korean
Statistical Information Service official website (accessed on
7 September 2017). Specifically, the unemployment rate
for those with less than a high school diploma has been
3.7, 2.7, 3.0 and 2.7% in 2000, 2005, 2010 and 2016,
respectively. On the other hand, the unemployment rates
for college graduates have been 6.3, 4.9, 4.6 and 3.9% in
2000, 2005, 2010 and 2016, respectively. Retrieved from
http://kosis.kr/eng/statisticsList/statisticsList_01List.
jsp?vwcd=MT_ETITLE&parentId=B#SubCont
624 © 2018 The Authors. International Journal of Social Welfare © 2018 Akademikerförbundet SSR (ASSR) and John Wiley & Sons Ltd
Recently, however, based on the recognition that
guaranteed universal access to social insurance pro-
grammes is necessary if all members of society are to
have a social safety net, various policy options have been
reviewed in Korea. As a result, the Durunuri programme
was initiated as a pilot project in 16 regions across the
country in February of 2012, and nationwide implemen-
tation followed several months later in July of 2012
(Kim, 2016; Yoo et al., 2013). Using empirical data on a
large nationally representative sample between the years
2010 and 2015, the present study examined the effective-
ness of a subsidy programme (i.e., the Durunuri) on
improving the level of social protection provided to soci-
ety’s most vulnerable population. The study employed a
quasi-experimental evaluation method to examine
whether the Durunuri programme had attained its pri-
mary goal of increasing low-wage workers’ subscription
rate to the national pension and employment insurance
programmes. Results from this study should provide a
more complete understanding of recent changes in
Korean social policies which can assist policymakers and
practitioners design and implement programmes that can
be used as a potential lever to improve outcomes for dis-
advantaged groups.
Background
Social investment perspective in Korea
As mentioned above, disparity in coverage of social
insurance programmes among workers is strongly
related to the ‘productivist’ nature of the Korean wel-
fare state where the goal of economic growth is pri-
oritised over other objectives when setting social
policy (Holliday, 2000; Kwon & Holliday, 2007).
The productivist welfare ideology is not unique to
Korea, rather it is a concept that was developed by
academics in comparative social policy to better
understand the welfare systems in East Asia’s high-
growth economies such as Korea, Japan, Hong Kong,
Singapore and Taiwan (Aspalter, 2006; Holliday,
2000). According to the East Asian welfare model,
productivist welfare states typically demonstrate high
spending on health and education, but tend to under-
develop their income maintenance programmes, rein-
forcing the position of productive elements in society
(Gough, 2000). A key feature of the productivist wel-
fare regime is, therefore, its targeted public invest-
ment approach in social affairs (i.e., the social
investment approach) to promote national and eco-
nomic development (Aspalter, 2006; Gough, 2000).
In the context of Korea, such ideology has been
manifested primarily through the development of
social insurance programmes the objectives of which
were to benefit productive sectors of the economy
with a high reliance on the private sector for service
delivery (Kwon & Holliday, 2007). Social insurance
Int J Soc Welfare 2018: 28: 63–76
 Effects of the Durunuri programme on social insurance coverage
Table 1. Social insurance subscription rate by employment status in 2015 (Unit: %).
Employment
insurance
Total 89.3
Standard employee 95.4
Non-standard employee 66.7
Fixed-term employee 87.0
Part-time employee 65.0
Daily employee 46.0
Temporary help service employee 90.7
Contract employee 94.6
Source: Ministry of Employment and Labor (Available at http://laborstat.moel.go.kr).
Note: The above information on social insurance subscription rate is based on the total number of legally ‘eligible’ workers for each social insurance pro-
gramme as estimated in the Survey on Labour Conditions by Employment Type.
programmes were designed to be contribution-based
rather than tax-based, and introduced in a series of
phases, first covering public sector employees and
workers in large enterprises before being extended to
the rest of the population. This targeted approach nat-
urally resulted in the concentration of benefits to the
skilled, male, industrial core workers (Peng, 2014).
However, following the Asian financial crisis of 1997
and more recently with the emergence of various socio-
economic problems such as increased income inequality,
low fertility rate and an ageing population, the politics of
welfare expansion has been noticeable (Choi, 2012). Spe-
cifically, in the past couple of decades, Korea has wit-
nessed an extension of coverage and benefits to social
security programmes as well as the implementation of
new social services including childcare programmes and
long-term care insurance for the elderly. Scholars have
been debating whether these recent expansions in social
policy since the early 2000s, which is not limited to
Korea but observed in other East Asian countries, imply
a diversion from the productivist welfare capitalism
approach (Aspalter, 2006; Choi, 2012; Kwon &
Holliday, 2007). It has become clear that the targets of
Korean social investment have expanded and shifted to
include more peripheral, marginalised and vulnerable
populations in hopes of addressing various socio-
economic issues (Choi, 2012; Peng, 2014). Implement-
ing policies that can increase both social and economic
sustainability has become more of an important policy
goal to the Korean government than ever before.
Korea’s recent adoption of MWP policies – EITC and
Durunuri social insurance programme – can be understood
from this perspective, where the social investment
approach is used not only to incentivise labour market par-
ticipation, but also to promote social inclusion and inter-
generational equity of the low-skilled population.
The Durunuri social insurance programme
In the past couple of decades, the need for universal
access to social security programmes has been elevated
Int J Soc Welfare 2018: 28: 63–76
© 2018 The Authors. International Journal of Social Welfare © 2018 Akademikerförbundet SSR (ASSR) and John Wiley & Sons Ltd
National National Industrial accident
health pension compensation
insurance insurance
88.4 88.7 97.6
97.9 97.8 98.0
55.5 52.7 96.4
92.6 88.5 98.3
59.6 56.4 94.2
8.5 9.0 97.4
91.3 90.7 96.7
92.5 89.8 98.3
in Korea. Specifically, following the Asian financial
crisis in 1997, Korea experienced unprecedented high
unemployment climbing up to 7–8% which, despite
falling quite rapidly, never returned to its previously
very low levels of 2–3% (Kwon & Holliday, 2007).
Measures taken to increase employment flexibility in
the labour market, which was a condition for the finan-
cial bailout programme of the International Monetary
Fund, resulted in a dramatic increase in the share of
non-standard workers who were generally not covered
by social security programmes (Shin, 2014). The
Korean government responded to such problems by
expanding the coverage of social insurance pro-
grammes (Kwon & Holliday, 2007; Shin, 2014). The
direction of such extensions moved from limited cover-
age of civil servants and employees in large firms to all
firms with fewer employees (Cho et al. 2008). As a
result, by 2003, each of the four major social insurance
programmes in Korea including the industrial accident
compensation insurance, the national health insurance,
the national pension, and the employment insurance
legally mandated all workers in workplaces with one or
more persons to be included in its statutory coverage
(Seo & Back, 2014).
Despite such efforts to expand coverage, exclusion
from actual coverage still exists today. Given that
these programmes protect only those members who
contribute to the system, workers who cannot afford
to pay their bills are not eligible to receive benefits.
Currently, the majority of Korean workers are gaining
benefits from social security; however, workers in
small-sized enterprises are at greater risk of not being
covered. According to reports on Labour Statistics by
Employment Types (available at http://laborstat.moel.
go.kr) in 2015, the enrolment rates of insured
workers were found to vary by employment status
(see Table 1). Among those who were eligible for
social insurance coverage, the overall average rate of
insured workers was above 88%, yet more than 95%
of those insured workers were standard (i.e., regular)
65
3
Park & Cho
Table 2. Social insurance subscription rate by size of firm in 2015 (Unit: %).
Employment insurance National health insurance
Fewer than 5 employees 69.4 68.9
5–29 employees 93.5 90.7
30–299 employees 97.4 97.2
More than 300 employees 96.8 99.7
Source: Ministry of Employment and Labor (Available at http://laborstat.moel.go.kr).
Note: The above information on social insurance subscription rate is based on the total number of legally ‘eligible’ workers for each social insurance pro-
gramme as estimated in the Survey on Labour Conditions by Employment Type.
employees. Regarding standard (i.e., regular) employees,
the enrolment rate in industrial accident pension
insurance was the highest (98%) and the rates for
other schemes were as follows: national health insur-
ance (97.9%), national pension (97.8%) and employ-
ment insurance (95.4%). However, for non-standard
employees, although the enrolment rate in industrial
accident compensation insurance was high (96.4%),
the rates for the other three remaining social insur-
ance programmes varied from 53 to 67%.
These numbers imply that there is still a substan-
tial portion of workers, especially among the non-
standard employees, who are legally eligible for
social insurance coverage but are not actually being
provided such services. It should also be noted that
the coverage rate decreases drastically when one
examines the total population of workers, including
those who are not legally eligible for benefits due to
age or employment-type restrictions. For example, in
2015, among all workers in Korea, the total enrol-
ment rate in the national pension was 82% for stand-
ard employees but only 36.9% for non-standard
employees, whereas for employment insurance it was
82.4% for standard employees versus 42.5% for non-
standard employees.2
In addition to employment type, company size
makes a huge difference, as employees in smaller
companies with lower wages are not accruing benefits
from the social security network (Kim, Go, & Song,
2015). Table 2 shows that, in 2015, eligible workers
in firms with fewer than five employees are discrimi-
nated against regarding coverage under social insur-
ance compared to large firms with 300 employees or
more. Only 69.4% of eligible workers in firms with
fewer than five employees were covered by the
employment insurance (as opposed to 96.8% in firms
with 300 employees or more), 68.9% by the national
health insurance (as opposed to 99.7% in firms with
300 employees or more) and 69.5% by the national
2
Information on social insurance programme coverage for
all workers (both eligible and ineligible) is available on the
Korean Statistical Information Service site. Retrieved 5
February 2018 from http://kosis.kr/statisticsList/statistic-
sListIndex.do?menuId=M_01_01&vwcd=MT_ZTITLE&-
parmTabId=M_01_01#SelectStatsBoxDiv.
646 © 2018 The
C Authors. International Journal of Social Welfare © 2018 Akademikerförbundet SSR (ASSR) and John Wiley & Sons Ltd
National pension Industrial accident compensation insurance
69.5 90.7
90.8 99.4
97.0 99.8
99.5 99.5
pension (as opposed to 99.5% in firms with 300
employees or more). As Shin (2014) argues, a large
gap exists between statutory coverage and actual cov-
erage among small firms because employers of
micro-enterprises are reluctant to pay social contribu-
tions for their employees in hopes of minimising
labour costs. This implies that the working poor are
rarely covered by social insurance programmes
through their workplace in the Korean context (Kim
& Lee, 2012).
The divide in working conditions between large
versus small- and medium-sized firms can be attrib-
uted to the segmentation of the Korean labour mar-
ket. That is, since the 1997 crisis, labour market
dualism has become more pronounced in Korea with
regular (i.e., standard) workers in large firms coexist-
ing with highly flexible irregular (i.e., non-standard)
workers in mostly small- and medium-sized firms
(Shin, 2014). The presence of strong union affiliation
in large firms with collective agreements and weak or
no union affiliation in small- and medium-sized firms
also contributes to this dualism. As a result, com-
pared to workers in small- and medium-sized firms,
employees of large firms tend to enjoy relatively high
wages, favourable employment conditions and full
social security coverage.
As a way to resolve the problem of low-skilled
workers’ exclusion from social insurance pro-
grammes, the Korean government launched the Duru-
nuri programme in July 2012 which subsidised up to
50% of both the employer and employee contribution
fees for the membership of the national pension and
employment insurance programmes. Providing incen-
tives for low-skilled workers and their employers to
enrol in the national pension and employment insur-
ance programmes was seen as a necessary policy
measure since it would allow the working poor to
maintain a minimum level of old-age security and
access unemployment benefits which could keep
them from falling into absolute poverty (Shin, 2014).
Given that social insurance subscription rates showed
huge variations with regard to wage level and work-
place size, the government restricted the eligibility of
the Durunuri programme to low-wage workers
employed in small businesses with fewer than 10
Int J Soc Welfare 2018: 28: 63–76
 Effects of the Durunuri programme on social insurance coverage
employees. In the first year of implementation, only
businesses with employees earning a salary of
1,250,000 KRW (equivalent to 1,165.66 US dollars
using 26 February 2018 exchange rate) per month or
below were supported, and the rate of financial aid
was differentiated up to a maximum of 50% subsidi-
sation in contribution fees. However, in January
2014, the programme broadened the target group by
increasing the salary threshold to include businesses
with employees earning a monthly salary of
1,300,000 KRW or below, while the rate of aid was
raised to a fixed rate of 50% in April 2013. The
monthly salary threshold continued to increase gradu-
ally over the years to the point that, in 2014, the eli-
gibility of monthly salary was increased slightly to
1,350,000 KRW and by January 2015 it became
1,400,000 KRW. In addition, in 2014, the subsidisa-
tion rate of contribution fees was no longer differenti-
ated and everyone eligible received 50%, whereas in
2016 the subsidisation rate became differentiated
again, with new members receiving 60% and existing
members receiving 40%. Most recently from January
2018, the new administration has drastically increased
benefits for the working poor by revising the monthly
salary threshold from 1,400,000 KRW to 1,900,000
KRW. Additionally, the subsidisation rate was
increased to 90% (or 80%) for new members
employed in businesses with fewer than five (or five
or more but fewer than 10) employees, but the subsi-
disation rate of 40% was maintained for existing
members.
In addition to the above-mentioned financial incen-
tives, the government took several measures to
increase participation, such as promoting public
awareness of the Durunuri programme and continu-
ously monitoring its take-up rate. A website was cre-
ated specifically for the Durunuri programme (http://
insurancesupport.or.kr), and to ensure that informa-
tion was easily accessible, online cartoons and guide-
books were produced. Both the National Pension
Service (http://www.nps.or.kr) and the Korea
Workers’ Compensation & Welfare Service (https://
www.kcomwel.or.kr) provide in-person site visits to
any eligible employee’s workplace inquiring about the
programme. Furthermore, these public corporations
cooperate with local governments and organisations
regularly by hosting contests or awarding businesses
that honour best practices in expanding social insurance
coverage of the Durunuri programme (Korea Workers’
Compensation & Welfare Services Press Release,
2015, 2017).
The Durunuri programme quickly became a major
policy in the Ministry of Employment and Labour
with an annual budget of roughly a quarter (i.e.,
25.2%) of the total annual budget for the entire
Ministry in 2016 (Kim, 2016). The budget grew
Int J Soc Welfare 2018: 28: 63–76
© 2018 The Authors. International Journal of Social Welfare © 2018 Akademikerförbundet SSR (ASSR) and John Wiley & Sons Ltd
Table 3. Number of workplaces and workers subscribed to the
National Pension and Employment Insurance programme: Among
workplaces with fewer than 10 employees during 2010 and 2014.
National pension Employment insurance
Year Workplaces Insured persons Workplaces Insured persons
2010 864,343 2,837,855 1,201,755 2,661,896
2011 929,441 2,995,726 1,299,837 2,889,863
2012 1,017,323 2,557,794 1,378,826 2,819,018
2013 1,105,452 3,386,619 1,512,104 2,947,433
2014 1,199,405 3,583,160 1,671,947 3,024,643
Source: National Pension Statistical Yearbook, by National Pension Serv-
ice (2010–2014). Yearly Statistics of Employment Insurance, by Ministry
of Employment and Labour (2010–2014).
annually from 0.2654 trillion KRW in 2012–0.5904
trillion KRW in 2015, and slightly decreased in 2016
– 0.5202 trillion KRW because of the differentiated
rate scheme between new and existing members. Of
course, the annual budget is expected to increase sig-
nificantly in 2018, given the revisions to both eligi-
bility and benefit levels.
In Table 3, among workplaces with fewer than 10
workers, the number of workplaces and workers sub-
scribed to the national pension and employment
insurance programmes is presented from 2010 to
2014. Given that the Durunuri programme was imple-
mented in 2012, it is important to see if membership
rates changed during the pre–post implementation
years. As is shown, both the number of workplaces
and workers covered by the two social insurance pro-
grammes increased following the implementation of
the Durunuri programme. However, it is difficult to
argue that such observed progress is due only to the
Durunuri programme since there have been consistent
efforts on part of the government to boost the rate of
subscription to social insurance programmes in gen-
eral, as well as other idiosyncratic factors that have
not been controlled for. Therefore, to measure the
effectiveness of the Durunuri programme, it is impor-
tant to separate the true effect of the Durunuri pro-
gramme from any natural increase of participants due
to other factors, such as timing effects.
Despite the expansion of the Durunuri social insur-
ance programme in recent years, extant literature
examining its effect has been limited (Choi & Cho,
2014; Yoo et al., 2016). That is, only a few studies
have discussed the Durunuri programme and the
effect of the Durunuri pilot project (Choi & Cho,
2014; Yoo et al., 2016). Specifically, Yoo et al.
(2016) evaluated the effects of the Durunuri pilot pro-
ject and the effects of the actual programme on a
worker’s probability to become employed as well as
join the employment insurance and national pension
programmes. Estimates using the difference-in-
differences (DID) method suggested that the pilot
67
5
Park & Cho
programme increased the number of the national
employment insurance subscribers by 2.4% and that
of the national pension subscribers by 3.1%, while it
failed to affect employment. They also analysed the
effects of the actual programme implemented from
July to December of 2012 (over 6 months) and found
that it had increased the number of employment
insurance subscribers by 2.68% and that of national
pension subscribers by 2.04%, while it had no effect
on employment. Choi & Cho (2014) compared the
national pension subscription rate of workers in small
businesses with fewer than 10 employees prior to the
Durunuri programme implementation and after imple-
mentation and found that subscription rates had
increased by 31.3% from July to December of 2012
as compared with the period between July and
December of 2011. Choi & Cho (2014) argued that
the findings suggested that the Durunuri programme
could have played an important role in encouraging
employers and workers of small businesses with
fewer than 10 employees to become new members of
the national pension programme despite acknowledg-
ing that the increase in subscription rate may not be
entirely due to Durunuri.
Although research conducted by Yoo et al. (2016)
and Choi & Cho (2014) provide interesting findings,
they both have limitations in that they analysed the
short-run effects (i.e., 6 months) of the Durunuri pro-
gramme. In addition, Choi & Cho (2014) was not able
to control for temporal effects or unobserved character-
istics that may have influenced one’s chances of enroll-
ing in the national pension programme since it simply
compared pre–post changes in the number of subscrib-
ers. In the present analyses, we attempt to overcome
the shortcomings of extant research by using recent
data and employing a rigorous methodology to control
for a number of confounding factors.
Research design
Model and method
This study utilised the DID method, a quasi-
experimental model, to analyse the effects of the
Durunuri social insurance programme. DID is com-
monly employed to assess the effectiveness of a pol-
icy by comparing the change in outcomes before and
after its intervention across a group affected by the
policy (i.e., treatment group) and a non-equivalent
comparison group not affected by the policy. There-
fore, in this study, changes in subscription status of
eligible Durunuri beneficiaries in the national pension
and employment insurance programmes are compared
with changes in subscription status of non-eligible
Durunuri beneficiaries. The disparities between the
two are considered as the effect of the Durunuri
programme.
68 © 2018 The Authors. International Journal of Social Welfare © 2018 Akademikerförbundet SSR (ASSR) and John Wiley & Sons Ltd
The DID method is able to measure the effect of
policy intervention effect precisely by controlling for
individual characteristics and other factors that are
correlated to the treatment group variable as well as
the outcome variable. Below, in equation (1), the
effect of the Durunuri programme is estimated using
logistic regression as follows:
!
X
2015
Yi 5a1b0 Treatedi 1b1 Treatedi 3 Dt
t52013
(1)
X
2015
1cXi 1 Dt 1Ei
t52011
The dependent variables used to estimate the effect
of the Durunuri programme are the subscription sta-
tus to the national pension programme and the
employment insurance programme – 1 if subscribed
and 0 otherwise. The dummy variable Treatedi takes
the value of 0 if the employee is ineligible for the
Durunuri programme, and 1 otherwise. That is, given
that only low-wage workers in workplaces with fewer
than 10 employees are eligible for the Durunuri pro-
gramme, workers in workplaces with more than 10
employees are treated as the comparison group in this
design. To construct an appropriate comparison
group, the present study relied on the definition of
small businesses/enterprises in the Korean context as
well as data availability. According to the Korean
Ministry of Small & Medium Enterprises and Start-
ups (www.mss.go.kr), small businesses are defined as
either microenterprises or small-sized enterprises
where the former includes workplaces with fewer
than 10 employees and the latter includes workplaces
with 10 or more but fewer than 50 employees.3 In
addition, information on workplace size was collected
as a categorical variable: (i) 1–4 persons; (ii) 5–9 per-
sons; (iii) 10–29 persons; (iv) 30–99 persons; (v)
100–299 persons; and (vi) 300 or more persons – in
the Economically Active Population Survey in
March, as described in the following Data section.
Therefore, the current study employed a comparison
group composed of workers in small-sized enterprises
with 10 or more but fewer than 30 employees. In
addition, as explained above, the monthly salary
threshold for Durunuri eligibility changed quite fre-
quently since its implementation in 2012, and there-
fore, individuals in both the treatment and
3
In January 2016, the Ministry of Small & Medium Enter-
prises and Startups (then called the Small & Medium Busi-
ness Administration) revised the definition of small-sized
enterprise to be an enterprise that has annual sales revenue
either less than 1 billion Korean won or between 1 and 12
billion Korean won. The definition of microenterprises
remained unchanged.
Int J Soc Welfare 2018: 28: 63–76
 Effects of the Durunuri programme on social insurance coverage
Table 4. Definition and measurement of variables.
Variable
Dependent variables Subscription status of national pension
Subscription status of employment insurance
Policy variables DID estimator of Durunuri
Group dummy
Control variables Gender
Age
Education level (omitted group: middle school or less)
Marital status
Log household income
Year (omitted group: 2010)
comparison groups were selected based on those
restrictions. The main variable of interest is the inter-
action term between the dummy variable indicating
treatment group status and the post-implementation
P individuals who are in the treatment group versus
year dummy variables, (Treatedi 3 2015 t52013 Dt ), which
is the DID estimator. P The post-implementation year
dummy variables ( 2015 t52013 D t ) take the value of 1
for the years 2013, 2014 and 2015, and 0 otherwise
(i.e., 2010, 2011 and 2012).4 In this model, b1 indi-
cates any change in the level of the outcome varia-
bles between the two groups following the
implementation of the Durunuri programme in years
2013, 2014 and 2015, respectively. Xi is a vector of
demographic characteristics including gender, age,
education level, marital status and household
income.P Lastly, we control for year fixed effects
(i.e., 2015
t52011 Dt ) to capture any economic, social or
policy factors that are common to a given year from
2010 to 2015. A detailed description of the variables
included in equation (1) is presented in Table 4.
Data
The data set used for this study is the Economically
Active Population Survey in March collected by the
Office of Statistics in Korea. This survey is con-
ducted to investigate Koreans’ economic activities
(i.e., employment, unemployment, labour force, etc.)
using a nationally representative sample and to pro-
vide basic data needed for the development of macro-
economic policy and research. The Economically
Active Population Survey has been conducted
4
The Economically Active Population Survey data are col-
lected in March of every year, and thus the 2012 data are
considered to reflect subscription rates prior to the imple-
mentation of the Durunuri programme (which was in July
2012).
Int J Soc Welfare 2018: 28: 63–76
© 2018 The Authors. International Journal of Social Welfare © 2018 Akademikerförbundet SSR (ASSR) and John Wiley & Sons Ltd
Measurement
Insured member 5 1, uninsured person 5 0
Insured member 5 1, uninsured person 5 0
After 1 year Group dummy 3 programme period (2013)
After 2 years Group dummy 3 programme period (2014)
After 3 years Group dummy 3 programme period (2015)
Comparison group Enterprises with fewer than 10 employees 5 1
Enterprises with more than 10 and
less than 30 employees 5 0
Male 5 1, Female 5 0
Continuous variable
High school 5 1, other 5 0
2-year college 5 1, other 5 0
4-year university or more 5 1, other 5 0
Married 5 1, not-married 5 0
Continuous variable (natural logarithm)
Dummy variable by each year
(2011, 2012, 2013, 2014, 2015)
annually every March since 1963 as a repeated cross-
sectional survey.
The final sample of this study can be divided into
those in the comparison group. Given that the Duru-
nuri programme’s target population is low-wage
workers, the study excluded individuals in the Eco-
nomically Active Population Survey who were
observed to earn more than the monthly income
threshold as prescribed by the government. Specifi-
cally, the low-wage threshold varied across the sam-
pling period: (i) those who earn a monthly income
less than 1,250,000 KRW (from 2010 to 2012); (ii)
those who earn a monthly income less than
1,300,000 KRW (from 2013 to 2014); and (iii) those
who earn a monthly income less than 1,400,000
KRW in 2015. Treatment group members included
low-wage workers from businesses with fewer than
10 employees between 2010 and 2015. The compari-
son group included low-wage workers in workplaces
with more than 10 and fewer than 30 employees. The
final sample included 39,410 individuals.
The characteristics of the study sample by year are
presented in Table 5. For example, in 2015, there
were 4,252 individuals in the treatment group (i.e.,
low-wage workers in workplaces with fewer than 10
employees) and 1,529 individuals in the comparison
group (i.e., low-wage workers in workplaces with 10
or more but fewer than 30 employees). The propor-
tion of females in the sample increased from 68.4%
in 2010 to 70.8% in 2015, and the mean age of sam-
ple individuals was 49.33 in 2015. The disproportion-
ately high representation of females in the study
sample, as compared with males, reflects the relative
disadvantage of women in the Korean labour market,
i.e., women are more likely than men to be employed
in small- and medium-sized firms as non-standard
69
7
Park & Cho

Table 5. Characteristics of the sample.

2010 2011 2012 2013 2014 2015

Variable People (%) People (%) People (%) People (%) People (%) People (%)

National pension Uninsured person 5,800 (73.7) 5,486 (75.2) 4,986 (75.4) 4,632 (75.7) 4,459 (77.7) 4,453 (77.0)
Insured member 2,071 (26.3) 1,806 (24.8) 1,626 (24.6) 1,484 (24.3) 1,279 (22.3) 1,328 (23.0)
Employment insurance Uninsured person 5,512 (70.1) 5,066 (69.6) 4,640 (70.3) 4,337 (71.0) 4,143 (72.3) 4,029 (69.8)
Insured member 2,347 (29.9) 2,213 (30.4) 1,959 (29.7) 1,772 (29.0) 1,588 (27.7) 1,747 (30.2)
Scale of Enterprise Fewer than 10 5,633 (71.6) 5,143 (70.5) 4,821 (72.9) 4,497 (73.5) 4,259 (74.2) 4,252 (73.6)
(Treatment group)
More than 10-less than 30 2,238 (28.4) 2,149 (29.5) 1,791 (27.1) 1,619 (26.5) 1,479 (25.8) 1,529 (26.4)
(Comparison group)
Gender Female 5,384 (68.4) 5,011 (68.7) 4,692 (71.0) 4,372 (71.5) 4,126 (71.9) 4,094 (70.8)
Male 2,487 (31.6) 2,281 (31.3) 1,920 (29.0) 1,744 (28.5) 1,612 (28.1) 1,687 (29.2)
Age Mean (Std. Dev) 45.31 (0.17) 46.46 (0.18) 47.29 (0.19) 48.53 (0.20) 49.30 (0.22) 49.33 (0.22)
Education Middle School or less 2,944 (37.4) 2,857 (39.2) 2,509 (39.0) 2,354 (38.5) 2,216 (38.6) 2,153 (37.2)
High school 3,497 (44.4) 3,124 (42.8) 2,939 (44.5) 2,630 (43.0) 2,471 (43.1) 2,569 (44.4)
2-year college 692 (8.79) 611 (8.38) 567 (8.58) 541 (8.85) 470 (8.19) 453 (7.84)
4-year university or more 738 (9.38) 700 (9.60) 597 (9.03) 591 (9.66) 581 (10.1) 606 (10.5)
Marital status Not married 3,020 (38.4) 2,755 (37.8) 2,521 (38.1) 2,296 (37.5) 2,256 (39.3) 2,281 (39.5)
Married 4,851 (61.6) 4,537 (62.2) 4,091 (61.9) 3,820 (62.5) 3,482 (60.7) 3,500 (60.5)
Household income Mean (Std.) 77.81 (0.35) 78.25 (0.37) 78.52 (0.40) 78.61 (0.44) 7792 (0.46) 84.59 (0.48)
N 7,871 7,292 6,612 6,116 5,738 5,781

workers. In 2015, the proportion of high school grad-
uates was the highest among sample individuals at
44.5%, followed by middle school or lower educated
people at 36.3%, university graduates at 11.1% and
college graduates at 8.0%, married people constituted
61.4% of the sample and the average monthly income
is reported to be 854,500 KRW; and finally, the over-
all rate of subscription to the national pension pro-
gramme and the employment insurance programme is
only 26.1% and 33.5%, respectively.
Results
Descriptive statistics: comparability between the
treatment and comparison groups
To explore the comparability of individuals in the
treatment versus the comparison groups, we statisti-
cally tested for differences across groups in terms of
a number of observed characteristics. Results from
the intergroup differences are presented in Table 6.
First, as is shown in Table 6, low-wage workers in
workplaces with fewer than 10 employees (i.e., treat-
ment group) appeared to be statistically significantly dif-
ferent from low-wage workers in workplaces with 10 or
more but fewer than 30 employees (i.e., first compari-
son group) in terms of all characteristics at the 5% sig-
nificance level. Specifically, the proportion of
individuals in the treatment group who participated in
the national pension programme was only 18.4%,
whereas the proportion of individuals in the comparison
group who participated in the national pension pro-
gramme was 43.6%. Similarly, the proportion of indi-
viduals in the treatment group who participated in the
employment insurance programme was 22.2%, whereas
780 © 2018 The Authors. International Journal of Social Welfare © 2018 Akademikerförbundet SSR (ASSR) and John Wiley & Sons Ltd
the proportion of participants in the comparison group,
at 49.1%, was more than double that rate. These esti-
mates indicate that the scale of business affects the
membership status of the national pension and employ-
ment insurance programmes, i.e., low-wage workers in
the treatment group were indeed disadvantaged in terms
of social insurance coverage as compared with low-
wage workers in the comparison group. These results
are consistent with Kim et al.’s (2015) argument stating
that most employees working for smaller businesses are
not protected by the social safety net in Korea.
Next, regarding gender, it appears that female
employees outnumbered male employees regardless
of the size of the company. Female workers com-
prised more than two-thirds of the treatment group
(69.9%) as well as of the comparison group (71.1%).
This is consistent with observations that, in Korea,
monthly income of female workers is lower than that
of male workers, and employment instability is rela-
tively higher. Individuals in the treatment group were
slightly younger than those in the comparison group.
Regarding education levels, high school graduates
constituted the highest proportion in both groups
(41.9–44.4%) followed by middle school or less
(37.5–39.8%). The proportion of individuals with ter-
tiary education levels did not vary much across the
two groups. Overall, the difference in education level
across groups was relatively small. Lastly, low-wage
workers in the treatment group were less likely to be
married and had lower average monthly income as
compared with those in the comparison group.
Finally, given the observed differences between
low-wage workers in workplaces with fewer than 10
employees (i.e., treatment group) versus low-wage
Int J Soc Welfare 2018: 28: 63–76
 Effects of the Durunuri programme on social insurance coverage

Table 6. Descriptive statistics of employees by inter-group difference analysis.

(A) (B)
Treatment group Comparison group
Fewer than 10 employees More than 10 and fewer than 30
employees

Variable Frequency (%) Frequency (%) v2 =t

(A) and (B)
p-Value

Outcome variables
National pension Uninsured person 23,352 (81.6) 6,464 (59.8) 2.0e103**
Insured member 5,253 (18.4) 4,341 (40.2)
Total 28,605 (100.0) 10,805 (100.0)
Employment insurance Uninsured person 22,241 (77.8) 5,486 (50.9) 2.7e103**
Insured member 6,341 (22.2) 5,285 (49.1)
Total 28,582 (100.0) 10,771 (100.0)
Control variables
Gender Female 20,000 (69.9) 7,679 (71.1) 4.9707*
Male 8,605 (30.1) 3,126 (28,9)
Total 28,605 (100.0) 10,805 (100.0)
Age Mean (Std. Dev) 47.17 (0.094) 48.48 (0.148) 7.3490**
Education Middle school or less 10,736 (37.5) 4,297 (39.8) 22.3568**
High school 12,705 (44.4) 4,525 (41.9)
2-year college 2,416 (8.5) 918 (8.5)
4-year university or more 2,748 (9.6) 1,065 (9.9)
Total 28,605 (100.0) 10,805 (100.0)
Marital status Single 11,313 (39.6) 3,816 (35.3) 59.3878**
Married 17,292 (60.5) 6,989 (64.7)
Total 28,605 (100.0) 10,805 (100.0)
Household income Mean (Std.) 77.31 (0.198) 83.99 (0.314) 17.8094**
No. of observations 28,605 10,805

*p<0.05; **p<0.01.

workers in workplaces with 10 or more but fewer
than 30 employees (i.e., the comparison group), we
attempted to quantify the comparability across these
groups by estimating one’s propensity to become eli-
gible for the Durunuri programme. That is, to further
examine whether individuals in the treatment group
are indeed comparable to those in the comparison
group in terms of observed characteristics, the non-
parametric kernel density estimate of one’s propensity
Figure 1. Non-parametric Kernel Density Estimates of the
propensity scores for being employed in a workplace with fewer
than 10 employees.
Note: The propensity scores are estimated by a logistic model controlling
for the following variables: gender, age, age 3 age, education level,
marital status, gender 3 marital status, income, and income 3 income.
In operating the logistic model, the balancing property is applied.
Int J Soc Welfare 2018: 28: 63–76
© 2018 The Authors. International Journal of Social Welfare © 2018 Akademikerförbundet SSR (ASSR) and John Wiley & Sons Ltd
to work in a workplace with fewer than 10 employ-
ees was estimated. The propensity scores were
obtained from logistic regression models controlling
for gender, age, education level, marital status, house-
hold income and a set of interactions and higher
order terms of variables included to satisfy the bal-
ancing property. Results are presented in Figure 1.
As is shown in Figure 1, the estimated propensity
scores for the treatment group is slightly higher than
those for the comparison group indicating that individ-
uals in the treatment group are more likely to be eligi-
ble for the Durunuri programme. However, despite
such small differences, the overlap in propensity scores
between the treatment group and the comparison group
demonstrates the existence of a wide area of common
support. The similarity in their estimated propensity
scores gives us a minimum assurance that although
individuals in the treatment and comparison groups
were found to be statistically significantly different
from each other in Table 6, comparison between the
two groups after controlling for such observed differen-
ces shall render meaningful results.
Descriptive statistics: outcome variable
Change in membership rates of those enrolled in the
national pension and employment insurance pro-
grammes by treatment group status are displayed in
9
71
Park & Cho

Figure 2. Change in the membership rate of the national pension and employment insurance programmes by treatment group status.
Note: The annual membership rate of the treatment group is calculated by dividing the number of low-wage workers in businesses with fewer
than 10 employees who are subscribed to the insurance programme in a given year by the total number of low-wage workers in businesses
with fewer than 10 employees in that given year. Likewise, the annual membership rate of the comparison group is calculated by dividing the
number of low-wage workers in businesses with more than 10 and fewer than 30 employees who are subscribed to the insurance programme
in a given year by the total number of low-wage workers in businesses with more than 10 and fewer than 30 employees in that given year.

Figure 2. As is shown, prior to the implementation of
the Durunuri programme during 2010 and 2012, the
membership rates for the national pension programme
had decreased slightly from 13.9% in 2010 to 13.3%
in 2012 for those in the treatment group but did not
vary much following implementation in 2013. Specif-
ically, the membership rates were 13.7% in 2013,
12.9% in 2014 and 13.0% in 2015. On the other
hand, the rate of enrolment in these social insurance
programmes for the comparison group displayed a
downward trend during this time period. The mem-
bership rates declined from 12.4% in 2010 to 11.3%
in 2012 and from 10.6% in 2013 to 10.0% in 2015.
Similarly, membership rates in the employment
insurance programme displayed divergent patterns
during the sampling time period for the treatment and
comparison group workers. For workers in the treat-
ment group, membership rates in the employment
insurance programme were slightly increasing prior
to the implementation of the Durunuri programme
from 15.4% in 2010 to 16.2% in 2012. This trend
did not change with the passage of the Durunuri pro-
gramme and the subscription rate continued to
increase from 16.3% in 2013 to 17.3% in 2015.
However, the rate of subscription for workers in the
comparison group actually declined during the same
time period from 14.5% in 2010 to 13.5% in 2012
and continued to drop following the implementation
of the Durunuri programme from 12.7% in 2013 to
11.8% in 2014 and 13.0% in 2015. Although the rate
of subscription increased for those in the comparison
group in 2015, it was still much lower than the rate
displayed by workers in the treatment group. These
raw differences across groups imply that the Durunuri
programme may have helped prevent low-wage
workers in the treatment group from losing coverage
10
72 © 2018 The Authors. International Journal of Social Welfare © 2018 Akademikerförbundet SSR (ASSR) and John Wiley & Sons Ltd
in both the national pension and employment insurance
programmes during the years following Durunuri
implementation. In the following section, we estimate
the DID estimator to examine whether the Durunuri
programme actually improved the social insurance cov-
erage rate of workers in the treatment group controlling
for observed differences.
Effects of Durunuri on the national pension and
employment insurance subscription rates
The results from estimating equation (1) using logis-
tic regressions are presented in Table 7. The odds
ratio of increasing the independent variable by one
unit is obtained by exponentiating the coefficient of
logistic regressions (i.e., exp(b)). The odds ratio of
the Durunuri variable in the current study is literally
the ratio of odds of two distinct states (i.e., treatment
vs. comparison group status) and can be interpreted
as follows: odds ratio of value 1 indicates that there
is no difference in the outcome variable between the
treatment and comparison groups; odds ratio greater
than 1 indicates that the treatment group has a higher
value in the outcome than the comparison group; and
odds ratio less than 1 indicates that the comparison
group has a higher value in outcome than the treat-
ment group.
As is presented in columns (1) and (2), the Durunuri
programme increased low-wage workers’ subscription
rates in both the national pension and employment
insurance programmes. Specifically, the odds ratio of
subscribing to the national pension and employment
insurance programmes for workers in the treatment
group were found to be as high as 16.5% and 16.0%,
respectively, 1 year after the implementation of the
Durunuri programme compared with those in the
Int J Soc Welfare 2018: 28: 63–76
 Effects of the Durunuri programme on social insurance coverage

Table 7. Effects of Durunuri on the national pension and employment insurance subscription rates.

(1) (2)

National pension Employment insurance

Variable Exp(b) z Exp(b) z

Durunuri After 1 year 1.165432 1.83 1.159607 1.89

After 2 years 1.266002 2.68 ** 1.17544 1.97*
After 3 years 1.292838 3 ** 1.243627 2.69**
Year (2010) 2011 0.869903 23.12** 0.997097 20.07
2012 0.87523 22.91** 0.967863 20.78
2013 0.775047 23.49** 0.835301 22.6**
2014 0.669062 25.23** 0.796126 23.15**
2015 0.559167 27.7** 0.734819 24.27**
Group dummy 0.272373 232.67** 0.258853 236.94**
Female 0.362793 230.73** 0.552123 220.42**
Age 0.965159 226.89** 0.991741 26.96**
Education level (middle school or less) High school 1.103024 2.59* 0.984536 20.46
2-year college 2.04899 13.09** 1.891734 12.59**
4-year university or more 1.55508 8.38** 1.233742 4.39**
Marital status (non-married) Married 1.579495 13.8** 1.268021 8.18**
Log household income 19.30118 54.14** 11.97684 55.37**
Constant 7.16E-06 247.08** 2.79E-05 249.6**
N 39,410 39,353
Log pseudolikelihood 215998.382 218697.333
Wald chi2 (15) 5982.60** 5578.60**
Pseudo R2 0.2686 0.2172

*p<0.05; **p<0.01.

comparison group. These estimates are marginally sig-
nificant at the 10% level. Following the first year of
implementation, however, both the effect size and sta-
tistical significance of the estimates increased. That is,
in the second year after implementation of the Duru-
nuri programme, the odds ratio of subscribing to the
national pension and employment insurance pro-
grammes for workers in the treatment group was esti-
mated to be as high as 26.6% and 17.5%, respectively,
compared with those in the comparison group. Simi-
larly, during the third year after implementation of the
Durunuri programme, the odds ratio of subscribing to
the national pension and employment insurance pro-
grammes for workers in the treatment group was esti-
mated to be as high as 29.2% and 24.3%, respectively,
compared with those in the comparison group. In addi-
tion, control variables such as gender, age, education
level, marital status and monthly income were all found
to significantly influence the subscription rate to the
social insurance programmes in expected directions.
That is, being male, younger, having higher levels of
education, being married and having higher income
levels were associated with greater likelihood of being
enrolled in both the national pension and employment
insurance programmes. In sum, results suggest that the
Durunuri programme has a positive impact on sub-
scription rates to the national pension and employment
insurance programmes, consistent with findings from
extant studies (Choi & Cho, 2014; Kim, 2014; Yoo
et al., 2016).
Int J Soc Welfare 2018: 28: 63–76
© 2018 The Authors. International Journal of Social Welfare © 2018 Akademikerförbundet SSR (ASSR) and John Wiley & Sons Ltd
Discussion and conclusion
The purpose of this study was to examine the effects
of the Durunuri programme in Korea, a MWP policy
that subsidises social insurance contributions of both
low-wage workers and their employers. The study
employed a treatment and comparison group strategy
in estimating the effects of the Durunuri programme on
the subscription to national pension and employment
insurance programmes. Findings revealed that the Dur-
unuri programme had a positive effect on increasing
low-wage workers’ social insurance coverage. The
major analytical results of this study were as follows:
First, our investigation confirmed that, compared
with low-wage workers in businesses with 10 or more
but fewer than 30 employees, low-wage workers in
businesses with fewer than 10 employees were found
to experience a 16.5, 26.6 and 29.2% increase in the
odds of being subscribed to the national pension pro-
gramme during the first, second and third year, respec-
tively, following implementation of the Durunuri
programme. Second, compared with low-wage workers
in businesses with 10 or more but fewer than 30
employees, low-wage workers in businesses with fewer
than 10 employees were found to experience a 16.0,
17.5 and 24.4% increase in the odds of being sub-
scribed to the employment insurance programme dur-
ing the first, second and third year, respectively,
following implementation of the Durunuri programme.
Such findings imply that Durunuri succeeded as a pol-
icy tool to increase the social insurance coverage of
73
11
Park & Cho
one of the most disadvantaged labour force groups in
Korea. Of course, it should be noted that such a con-
clusion is based on the assumption that eligibility rules
were met, and therefore, the Durunuri programme ben-
efits were provided to the truly deserving low-wage
workers who are the program target population. How-
ever, given some recent reports of employers underre-
porting employees’ salaries to reduce the financial
burden of social insurance contribution fees (particu-
larly national pension contributions), one cannot pre-
clude the possibility that Durunuri programme benefits
may have been extended to workers with actual wages
higher than stipulated by eligibility rules (Park, 2011).
From a practical viewpoint, additional efforts to accu-
rately identify the target population should be neces-
sary. It should also be important to verify to what
extent employers have illegally benefited from the Dur-
unuri programme in future research.
There are a couple of key limitations of our study.
First, the data set used in the present study, Economi-
cally Active Population Survey, is a repeated cross-
sectional survey which collects information on indi-
vidual economic activity every year. Since individu-
als are not followed across time, it is impossible to
estimate change in behaviour relative to a certain
point in time. This means that we cannot examine
whether an individual joined the national pension and
employment insurance programmes following the
passage of the Durunuri programme, nor can we
investigate individual patterns of moving in and out
of such programmes. In addition, although we find
strong evidence relating the Durunuri programme and
an increase in social insurance coverage, we are cau-
tious to make strong causal inferences. This is due to
the limited number of covariates in our data sets
which we could control for; therefore, differences in
both observable and unobservable characteristics may
remain between the treatment and comparison groups.
However, as we show in Figure 1, there is consider-
able overlap in propensity scores between the two
groups indicating that our evaluation strategy is valid.
It would be useful for future research to employ addi-
tional rigorous evaluation methods to confirm our
finding and move us closer to understanding the
causal effect of the Durunuri programme.
Given the economic context of Korea, a nation
with relatively low rates of social insurance coverage
for its working poor, findings from the present study
indicate that government subsidisation of social insur-
ance programmes may be an effective tool to reduce
the low-wage workers’ risks of falling into absolute
poverty (Kim, 2016). That is, from a theoretical
standpoint, these results confirm that an inclusive
approach to social investment, including not only the
productive sector but also vulnerable populations,
produces favourable results. Specifically, as
12
74 © 2018 The Authors. International Journal of Social Welfare © 2018 Akademikerförbundet SSR (ASSR) and John Wiley & Sons Ltd
beneficiaries of the Durunuri programme, the low-
skilled shall become eligible for benefits during peri-
ods of unemployment and receive pension funds in
old age. Given that the majority of low-wage employ-
ees in small businesses tend to be non-standard work-
ers with insecure jobs, being enrolled in these types
of social security programmes should be vital for
their current and future well-being. Research has
extensively documented that social insurance cover-
age is important for building a social security net for
the poor in Korea. For example, according to Seok
(2010), the poverty rate of elderly Korean households
declined by 23.3% after receiving public pensions.
Kang (2011) also found that the national pension pro-
gramme had a large and significant effect on reducing
poverty but that the magnitude of the effects varied
by age group 2 workers in their 40s were observed to
experience larger effects than those in their 50s and
30s. Given the recent expansions in the Durunuri pro-
gramme to include a larger population of the low-
skilled, future research should focus on examining
the long-term sustainability of this programme in
terms of consolidating work and welfare, thereby cre-
ating a balance between social and economic goals.
Next, results from the study raise questions around
the issue of equity in eligibility for the Durunuri pro-
gramme. As is explained, the Durunuri programme
has two strict eligibility requirements: the monthly
salary of individuals and workplace size. Use of
means-tested schemes in providing welfare benefits is
quite common in Korea, and therefore, the first
requirement did not raise much controversy. How-
ever, the validity of the second requirement, work-
place size, has sparked controversy from its very
inception, and the question of whether restricting
benefits only to low-wage workers in firms below a
certain size is reasonable has been raised since. That
is, initially, in 2011, during the early stages of plan-
ning, the government had proposed to impose a
much more limited provision criterion of the
Durunuri programme related to concerns about high
costs of the programme. Specifically, the government
had planned to provide subsidies only to low-wage
employees in firms with fewer than five employees
and to cap the maximum benefit level to one third
(rather than half) of contribution fees (Kim, 2016).
However, during the 2012 congressional budget
review process, both the level of benefit and target
groups were expanded, and additional funds were
allotted to implement the current programme (Kim,
2016). To this day, debates related to equity concerns
about the workplace size requirement continue. For
instance, on 19 June 2017, a legislative initiative
requesting to eliminate restrictions based on work-
place size was submitted to Congress (Lee, 2017).
Int J Soc Welfare 2018: 28: 63–76
 Effects of the Durunuri programme on social insurance coverage
It should also be noted that there is the potential
for this debate to go beyond the issue of equity. In
fact, the result that the Durunuri programme may be
an important policy tool to enrol the low-skilled pop-
ulation in social insurance schemes implies that there
may be additional indirect economic benefits to the
programme which were not considered during the
process of planning. For example, since receiving
social security benefits has been shown to reduce
poverty rates (Kang, 2011; Seok, 2010), one may
speculate about indirect benefits associated with
reduced spending on social programmes for the poor.
Of course, a thorough investigation on both the short-
and long-run effects of the Durunuri programme on
poverty reduction would be necessary to draw any
definitive conclusion.
Finally, from a policy perspective, the inclusion of
the national health insurance programme in the
Durunuri benefits may be warranted. As is shown in
Table 2, roughly a third of workers in small workplaces
(i.e., fewer than five employees) are not subscribed to
three (i.e., employment insurance, national health insur-
ance and national pension) among the four most typical
social insurance programmes in Korea. Given that the
Durunuri programme targets only the national pension
and employment insurance programmes, low-wage
employees are left exposed to risks related to illness
and health problems without any social protection. This
is even more problematic given the recent rising trend
in the cost of outpatient care per visit for the bottom
quintile income group in Korea.5 Specifically, although
the average cost of outpatient care per visit rose by
6.7% from 2008 to 2013, the increase was 13.4% for
the bottom quintile income group whereas it was
21.3% for the top quintile income group. Not surpris-
ingly, it appears that the proportion of medical spend-
ing in Korean household expenditure data has been
increasing since the early 2000s. For example, medical
care expenditure accounts for 6.8% of the average
monthly household expenditure in 2015, an increase
from 5.8% in 2003 (Yoo, 2016). These statistics indi-
cate that without proper governmental support or inter-
vention, low-income households are at greater risk of
falling into poverty because of health reasons. As such,
the suggestion to include the national health insurance
programme in the Durunuri subsidies has been made
by other scholars (Kim, 2012).
Currently, the Moon administration (2017–present)
has reportedly been pursuing the inclusion of the
national health insurance programme in the Durunuri
subsidies, yet the exact level of coverage and timing
of initiation have yet to be declared (Yonhapnews,
5
Information on costs related to outpatient care is available at
http://kosis.kr/statHtml/statHtml.do?orgId=331&tblId=DT_
12345N_045&conn_path=I3 (accessed on 18 January 2017).
Int J Soc Welfare 2018: 28: 63–76
© 2018 The Authors. International Journal of Social Welfare © 2018 Akademikerförbundet SSR (ASSR) and John Wiley & Sons Ltd
2017). Given that the expansion of the programme
will incur additional costs, more research as well as a
rigorous cost-benefit analysis should be conducted for
the creation of a sustainable programme (Kim, 2016).
If current considerations to expand the coverage of
the Durunuri programme entail a need to contain pro-
gramme costs, possible options may include provid-
ing differential assistance by one’s level of monthly
salary as well as taking into consideration the number
of working hours (Kim, 2016; Lee et al., 2014). Such
institutional changes would not only improve the effi-
ciency of the programme but could also maximise
effectiveness by reducing issues of moral hazard.
Acknowledgement
The work was supported by a grant from the National
Research Foundation of Korea funded by the Ministry
of Education in the Republic of Korea (NRF-
2016S1A3A2924832).
References
Aspalter, C. (2006). The East Asian welfare model. Interna-
tional Journal of Social Welfare, 15(4), 290–301.
Bargain, O., Caliendo, M., Haan, P., & Orsini, K. (2010). Mak-
ing work pay in a rationed labor market. Journal of Popula-
tion Economics, 23(1), 3232351.
Cho, J.-M., Kim, G.-S., & Kwon, T.-H. (2008). Employment
problems with irregular workers in Korea: A critical approach
to government policy. Pacific Affairs, 81(3), 407–426.
Choi, O. G., & Cho, Y. E. (2014). Current status and future
tasks of national pension social insurance support system,
Journal of Critical Social Policy, 45, 304–339 (in Korean).
Choi, Y. J. (2012). End of the era of productivist welfare capi-
talism? Diverging welfare regimes in East Asia. Asian Jour-
nal of Social Science, 40, 275–294.
Dagsvik, J. K., Jia, Z., Orsini, K., & Van Camp, G. (2011). Sub-
sidies on low-skilled workers’ social security contributions:
The case of Belgium. Empirical Economics, 40(3), 779–806.
Gough, I. (2000). Welfare regimes in East Asia and Europe.
Paper delivered at the Annual World Bank Conference on
Development Economics Europe 2000. Paris, 27 June.
Holliday, I. (2000). Productivist welfare capitalism: Social pol-
icy in East Asia. Political Studies, 48(4), 706–723.
Kang, S. H. (2011). The estimate of non take-up of National
Pension and the effect of National Pension on poverty
relief. Korean Journal of Public Finance, 4(2), 89–121
(in Korean).
Kim, J. (2016). Analysis on the Legislative Impact of Durunuri
Insurance Support System. Report on Legislative Impact
Analysis, 19, Publication No. 31–9735043-000143-14,
National Assembly Research service (in Korean).
Kim, J. J. & Lee, H. W. (2012). A study on linking EITC and
social insurance subsidy Programme. Research Report, Pub-
lication No. 978-89-8191-602-2, Korea Institute of Public
Finance (in Korean). Retrieved from http://www.kipf.re.kr/
Publication/B/%EA%B7%BC%EB%A1%9C%EC%9E%
A5%EB%A0%A4%EC%84%B8%EC%A0%9C(EITC)%
EC%99%80-%EB%91%90%EB%A3%A8%EB%88%84%
EB%A6%AC-%EC%82%AC%ED%9A%8C%EB%B3%B4%
ED%97%98-%EC%A7%80%EC%9B%90%EC%82%AC%
EC%97%85-%EC%97%B0%EA%B3%84%EB%B0%A9%
EC%95%88/521114
75
13
Park & Cho
Kim, T. H. (2014). A study on the blind spot of the social insur-
ance: Focused on Durunuri social insurance subsidy pro-
gramme. Kyungpook National University Law Journal, 46,
431–456 (in Korean).
Kim, Y. S., Go, J. S., & Song, A. J. (2015). Study on regional
employment enhancement plan through expansion of Durunuri
social insurance programme. Journal of the Korean Urban
Management Association, 28(3), 269–284 (in Korean).
Koh, Y. (2015). The social safety net in Korea: From welfare to
workfare. In: J. K. Scholz (Ed.), Social policies in an age of aus-
terity: A comparative analysis of the US and Korea (pp. 44–79).
Cheltenham, United Kingdom: Edward Elgar Publishing.
Korea Workers’ Compensation & Welfare Services Press
Release. (2015, July). Making Korea a nation with no social
insurance blind spot. Retrieved from https://www.kcomwel.
or.kr (in Korean).
Korea Workers’ Compensation & Welfare Services Press
Release. (2017, December). Korea Workers’ Compensation
& Welfare Services, selecting businesses with best practices
for encouraging social insurance programme participation.
Retrieved from https://www.kcomwel.or.kr (in Korean)
Kwon, S. M. & Holliday, I. (2007). The Korean welfare state: A
paradox of expansion in an era of globalization and economic
crisis. International Journal of Social Welfare, 16, 242–248.
Lee, B. H., Jang, J. E., Kim, H. W., Lee, S. G., Shim, G. B., &
Kang, B. G. (2014). Analysis on exclusion from social insur-
ance coverage and mid- to long-term recommendations.
Research Project in the Ministry of Employment and Labor,
Korea Labor Institute (in Korean). Retrieved from http://
www.prism.go.kr/homepage/researchCommon/retrieveRe-
searchDetailPopup.do?research_id= 1490000-201500053
Lee, Y.-S. (2017, June). Congressman Yoon Soh-Ha, proposes
to amend the social security law to eliminate the workplace
size criteria of the Durunuri programme. www. DailyMedi-
pharm.com (in Korean). Retrieved from http://www.daily-
medipharm.com/news/articleView.html?idxno=35444
Marx, I., Vanhille, J., & Verbist, G. (2012). Combating in-work
poverty in Continental Europe: An investigation using the
Belgian case. Journal of Social Policy, 41(1), 19–41.
76
14 © 2018 The Authors. International Journal of Social Welfare © 2018 Akademikerförbundet SSR (ASSR) and John Wiley & Sons Ltd
Park, H. D. (2011, September). Large companies under-paying
employee pension contributions. . .under-reporting Salary up to
40%. KyunghyangBiz (in Korean). Retrieved from http://biz.
khan.co.kr/khan_art_view.html?artid = 201109190250555
& code = 920401
Peng, I. (2014). The social protection floor and the ‘new’ social
investment policies in Japan and South Korea. Global Social
Policy, 14(3), 389–405.
Seo, J.-H. & Back, S.-H. (2014). Analysis of changes in the
social insurance legislation and coverage: With the perspec-
tive of legal exclusion. Quarterly Journal of Labor Policy,
14(3), 37–77 (in Korean).
Seok, J. E. (2010). Impacts on the effectiveness and the effi-
ciency of poverty alleviation of Public Pension and Basic
Old-Aged Pension: Seoul metropolitan city. Social Welfare
Policy, 37(3), 193–214 (in Korean).
Shin, D. M. (2014). Towards dynamic social security in South
Korea. International Social Security Review, 67, 75–104.
Van der Linden, B. (2016). Do in-work benefits work for low-
skilled workers? IZA World of Labor, 1–10.
Yonhapnews. (2017, July). Low wage workers get subsidy for
national health insurance programme. . .Expansion of the
Durunuri Programme. Yonhapnews (in Korean). Retrieved
from http://www.yonhapnews.co.kr/bulletin/2017/07/21/02
00000000AKR20170721134600017.HTML?input=1195m
Yoo, G. J., Kang, C. H., & Choi, P. U. (2013). A performance
evaluation on a social insurance subsidy programme (Duru-
nuri Programme). Research Report, Series No. 2013-02,
Korea Development Institute, (in Korean). Retrieved from
http://www.kdi.re.kr/research/subjects_view.jsp?pub_no=13562
Yoo, G. J., Kang, C. H., & Choi, P. U. (2016). Measuring the
effects of a social insurance subsidy programme (Duru-Nuri
Programme): an application of programme evaluation meth-
ods. The Korean Economic Review, 64(1), 732106 (in
Korean).
Yoo, K. J. (2016). Social Indicators in 2015. Social Indicators
in Korea, 37, Publication No. 11–1240000-000039-0, Statis-
tics Korea (in Korean).
Int J Soc Welfare 2018: 28: 63–76