A review of international experience

Matching Contributions and 
Compliance in Korea’s 
National Pension Program
Hyungpyo Moon
To encourage compliance with its national pension program, the government of the Repub-
lic of Korea has been providing matching subsidies to farmers and fishers of up to half of 
their contribution since 1995. Regression estimates indicate that subsidized groups were 
more than 10 percentage points more likely to contribute than other self-employed workers, 
after controlling for other variables, and that differences in contribution rates were larger 
among low-income workers. The results imply that policy makers can expect contributions 
to the national pension system to rise if similar subsidies are provided to other groups of 
workers with low incomes.
E
normous changes are expected in the Republic of Korea as a result of the rapid aging 
of the population and the extremely low birth rate. The number of people age 65 and 
older will reach 16.2 million in 2050, 3.7 times the 4.4 million people this age in 2005, 
according to projections by the National Statistical Office (2006). A rapidly aging popula-
tion will increase the importance of old-age income security and significantly affect public 
pension finance. Korea thus now faces two major policy challenges: enhancing the role of 
the old-age income security system as early as possible and at the same time improving the 
financial sustainability of the public pension system.
Although the national pension became applicable to the entire population in 1999, 
more than a third of all workers are not contributing. A majority of these nonpayers are 
poor or low-income people with insecure employment, such as atypical workers, the self-
employed, and small business owners. Many of them will have difficulty securing pension 
rights after retirement. Even if they are entitled to pensions, the amount of their benefits 
may not be enough to support them. Extensive noncompliance increases the risk that 
many people will fall into poverty in old age.
Recognizing the problems of narrow coverage of the national pension (“pension 
blind spots”), the government has tried to expand pension coverage and compliance of 
groups in which contributions are low. In addition to building up the administrative abil-
ity to detect the actual income of informal workers and the self-employed, the govern-
ment is discussing more radical policy measures. One option is to encourage people who 
are not currently covered to participate by subsidizing their pension contributions. The 
main purpose of the contribution subsidy is to prevent old-age poverty rather than provid-
ing ex post financial support after people become poor late in life.
Subsidizing contributions to the national pension is not an entirely new idea in 
Korea. The government has been partially subsidizing the contributions of farmers and 

162 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
fishers to encourage their participation by reducing the contribution burden since com-
pulsory coverage of the national pension was extended to all residents in rural areas in 
1995. Has the contribution subsidy increased pension participation of farmers and fish-
ers? This chapter examines the empirical evidence on the effectiveness of the contribution 
subsidy by comparing the compliance behavior of subsidized farmers and fishers with 
behavior of nonsubsidized self-employed workers.
The chapter is organized as follows. The first section briefly reviews the narrow cov-
erage of the national pension scheme in Korea. The second section examines the compli-
ance behavior of the insured. The third section empirically investigates the effectiveness 
of the contribution subsidy on increasing compliance with the scheme. The last section 
summarizes the chapter’s findings.
Coverage of the National Pension
The Korean national pension is a traditional contributory defined benefit pension plan 
(box 8.1). General coverage of the national pension has grown rapidly since the scheme 
was first introduced in 1988 (figure 8.1). By 1999, it became a nominally universal pen-
sion system, after all working-age Koreans were obliged to contribute to the scheme. The 
scheme covers all resident citizens of Korea age 18–59, regardless of their income. About 
BOX 8.1  The National Pension Scheme in Korea
The national pension, as a defi ned benefi t scheme which was fi rst introduced in 1988, is 
designed to function as an income protection system against a wide range of social risks includ-
ing old age, disability, and death. All residents in Korea from 18 to less than 60 years of age, 
regardless of their income, are legally covered under the scheme. 
The basic pension amount, which is the basis for the calculation of old-age, disability, and 
survivor pension, is composed of earnings-related and redistributive components. The basic 
pension amount is in direct proportion to the insured period, but not to the insured person’s 
income. The current income replacement rate of old-age pension of the insured person with 
median income and a 40-year insured period is approximately 50 percent. The replacement rate 
is being reduced by 0.5 percentage points annually until reaching 40 percent in 2028. A price 
indexing system is adopted to maintain the real value of the pension amount. At present, the nor-
mal age eligible for an old-age pension is 60, but it shall be gradually increased to 65 by 2033. 
The expenditure required for payment of benefi ts is mainly fi nanced from contributions paid by 
insured persons and their employers. The contribution rate was set low at the initial stage of the 
scheme (3 percent in 1988) and has been gradually increased to the current 9 percent. It will be 
adjusted according to the fi nancial recalculation planned to be conducted every fi ve years. The 
government’s fi nancial support is temporarily provided for some portion of contributions paid by 
farmers and fi shers for the purpose of alleviating the fi nancial burden on them.
SOURCE: National Pension Service website (http://english.nps.or.kr).

8.  MATCHING CONTRIBUTIONS AND COMPLIANCE IN KOREA’S NATIONAL PENSION PROGRAM 
163
92 percent of the workforce age 18–59 is currently covered by public pension programs. 
The national pension covers 86 percent of the workforce; the other 6 percent are covered 
by pension schemes for civil servants, military personnel, and private school teachers.
People insured by the national pension scheme are divided into two main groups: 
people who are insured through their workplace and people who are insured individually. 
All employers are required to provide workplace-based coverage if they have even a single 
employee. The contributions of people covered by workplace-based insurance are shared 
equally by the employer and the employee. Employers handle the acquisition of pension 
rights, the loss of insured status, and the payment of contributions. All self-employed 
workers are individually insured. They are supposed to pay both halves of their contribu-
tion and report their income status themselves.
The category of individually insured people includes those who are exempted from 
making contributions. Where a mandatorily insured person has temporary difficulty mak-
ing a contribution because of termination of business, loss of employment, or a calamity 
or accident, he or she can be exempted from doing so during the relevant period without 
losing coverage. Once the person earns income again, he or she should then report to the 
National Pension Service and resume making contributions. According to official statis-
tics, 5 million people were exempted from making contributions in 2009, 27.1 percent of 
all insured people and 58.2 percent of all individually insured people (table 8.1).
Among income reporters, a large number of insured people remain delinquent. As 
of 2009, the system collected 92 percent of the amount it should have collected (from 
both workplace-based and individually insured workers), but only 62 percent of all indi-
vidually insured people made their contributions (National Pension Service 2009). About 
half of the people who failed to pay contributions are long-term delinquents with more 
than two years of deferral.
FIGURE 8.1  Coverage of Korea’s national pension system, 1988–2009
2
4
6
8
10
12
14
16
18
20
1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 
millions
workplace-based insured
individually insured
workplaces with
10 or more
employees 
rural areas, 
farmers and
fishers
urban self-employed,
workplaces with < 5
employees 
0
workplaces
with 5 or
more
employees 
workplaces
with 1 or more
employees 
SOURCE: National Pension Service website (http://english.nps.or.kr).

164 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
TABLE 8.1  Number of people insured by Korea’s national pension, 2009
Type of insured 
Number covered (millions)
Proportion covered (%)
Workplace based
9.87
53.0
Individually insured
8.68
47.0
Income reporter
3.63
19.9
Contribution exemption
5.05
27.1
SOURCE: National Pension Service 2009.
Extensive noncompliance raises the risk of falling into poverty late in life. Such 
risks are evident in cases of exemption from contribution payment. Almost 90 percent of 
the people exempted from payment were poor, and more than three-fourths were jobless 
(table 8.2). The number of unemployed among the exempted from contribution payments 
is much higher than the national unemployment rate (3.6 percent in 2009). A direct com-
parison may not be fair, however, given that the number of insured but exempted from 
payment includes the economically inactive population (who paid contributions while 
employed) as well as the unemployed. Nevertheless, the large gap between the two num-
bers implies the limited capability of the government to gather accurate information on 
the personal income of income earners. The majority of informal sector workers and self-
employed small business owners evade income reporting or underreport their incomes in 
order to avoid contributing to the pension plan. Their actual incomes are very difficult to 
measure because of the poor administrative capability of the pension system. As a result, 
although they are engaged in economic activities, many informal workers and small busi-
ness owners are exempt from paying contributions to the system.
TABLE 8.2  Reasons for exemption from contributing to Korea’s national pension plan, 2009
Reason for exemption
Number of exemptions (thousands)
Percentage of total
Jobless
3,817
75.6
Business closure
433
8.6
Student
278
5.5
Temporary leave
92
1.8
Livelihood difficulties
7
0.1
Other
a
422
8.4
Total
5,052
100.0
SOURCE: National Pension Service 2009.
A. Includes military personnel, inmates, missing people, hospital patients, and others.

8.  MATCHING CONTRIBUTIONS AND COMPLIANCE IN KOREA’S NATIONAL PENSION PROGRAM 
165
Who Contributes to the National Pension System and 
Who Does Not? 
This section examines the characteristics of people who contribute to the national pension 
and the reasons why people do not contribute. It is based on data from the 2008 Korea 
Welfare Panel Study, the largest source of panel data in Korea. The survey, which has fol-
lowed more than 7,000 households since 2006, includes farming and fishing households, 
which makes it suitable for this study.
1
 This analysis extracts 4,115 individuals age 18–59, 
the target age of the national pension, from the 2008 data.
About a third (31 percent) of the population in the sample did not contribute to the 
national pension (table 8.3). Among people insured through their workplace, 99.8 per-
cent contributed. In contrast, just 34.4 percent of individually insured people did so, 
mainly because 57 percent of all individually insured people were exempt. However, even 
among people who reported their income, the delinquency rate was almost 20 percent, 
much higher than the rate for people insured at their workplace.
More than half of respondents classified as exempt engage in economic activities, 
indicating the seriousness of “blind-spot” problems arising from poor administrative 
capability in gathering accurate information on income status. Among the individually 
insured, 33.4 percent are actually wage workers, the majority of whom are legally entitled 
to workplace-based insurance. These workers remained individually insured because their 
status reports were not submitted properly. 
The compliance rate of the individually insured shows considerable differences by 
occupational status. The noncompliance rate of wage workers averaged 72.2 percent—1.7 
times the 41.5 percent noncompliance rate among the self-employed. The noncompliance 
TABLE 8.3  Number of people covered by Korea’s national pension plan, by occupational and 
participation status, 2007
Category
Total  
Non–wage worker
Wage worker
Other
a
Sub-
total
Employer
Self- 
employed
Sub-
total
Regular 
worker
Temporary 
worker
Daily 
laborer
Workplace based
2,156
Contribution
2,152
62
57
5
2,081
1,888
171
22
9
Noncontribution
5
1
1
0
4
4
0
0
0
Individually insured
1,959
Contribution
674 406
129
277
182
92
29
62
85
Noncontribution
1,285 288
64
225
474
161
136
176
523
Exemption
1,111
195
46
149
412
146
115
151
504
Delinquent
169
92
18
75
60
15
20
25
17
SOURCE: Author, based on data from the Korea Welfare Panel Study 2008.
A. Includes unpaid family workers, unemployed, and economically nonactive population.

166 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
rate of wage workers was 63.6 percent for regular workers, 74.1 percent for daily labor-
ers, and 82.7 percent for temporary workers. These data suggest that nonregular workers 
should be considered the primary target in developing policy measures to reduce pension 
blind spots.
Table 8.4 shows the characteristics of individually insured and uninsured people 
(exempt payers and delinquents). The characteristic that stands out most is that the aver-
age disposable income of individually uninsured people is more than 20 percent lower 
than that of individually insured people. Also, as household incomes decrease, the non-
compliance rate increases rapidly, meaning that economic status is a significant determi-
nant of participation in the pension system. Individually uninsured people tend to have 
a higher education level than insured people, and younger people have lower compliance 
rates than older people (the noncompliance rate of people in their 30s or younger is as 
high as 95 percent). The compliance rate of women is about 20 percent lower than that of 
men, meaning there are bigger blind spots among women.
A regression analysis using the probit model was conducted in order to examine the 
effects of individual characteristics on compliance behavior. The results show the estima-
tion of the compliance probability of 1,993 individually insured people (table 8.5). As 
expected, the compliance probability of individually insured people is highly sensitive 
to variables such as sex, age, and income. The probability of compliance appears to rise 
significantly with income and age; it is also higher for men than women. The estimated 
coefficients on educational background are negative and statistically insignificant, indicat-
ing that education has little impact on compliance.
The compliance probability of individually insured people varies significantly with 
occupational status. The compliance probability of daily laborers is not statistically sig-
nificantly different from that of temporary workers (whose probability of compliance is 
17.4 percent). In contrast, the probability of compliance increases by 18.4 percentage 
points for regular workers and by 31.3 percentage points for the self-employed. Even after 
controlling for individual characteristics, among individually insured people the compli-
ance probability of wage workers and the self-employed shows considerable differences. 
The compliance probability of temporary and daily laborers is particularly low.
How Effective Has the Government’s Contribution Subsidy to 
Farmers and Fishers Been?
Recognizing that expanding coverage of the national pension is essential to preparing for 
a rapidly aging society, the government has made efforts to reduce the blind spots in the 
national pension. As part of such efforts, the Ministry of Health and Welfare recently 
undertook a review of a matching contribution subsidy that would provide low-income 
individually insured people with a subsidy that covers up to half of their contributions.
2
 
This program may serve as an incentive for low-income workers.
The government has been subsidizing monthly contributions by farmers and fishers 
to reduce the burden of pension contribution and encourage their participation ever since 
the national pension extended its compulsory coverage to all rural residents in 1995.
3
 
The subsidy plan is financed by revenues from an earmarked tax in the Special Accounts 

8.  MATCHING CONTRIBUTIONS AND COMPLIANCE IN KOREA’S NATIONAL PENSION PROGRAM 
167
TABLE 8.4  Demographic characteristics of individually insured and noninsured people in Korea’s 
national pension plan
Category
Total
Individually 
insured
Noninsured
No. of 
samples
%
No. of 
samples
%
No. of 
samples
%
Total
1,959
100.0
674
100.0
1,285
100.0
Age
18–30
205
10.5
11
1.6
194
15.1
31–40
684
34.9
166
24.6
519
40.4
41–50
650
33.2
264
39.2
385
30.0
51–59
420
21.5
233
34.6
187
14.6
Average
42.4
46.5
40.3
Education
Uneducated, elementary, 
middle school
329
16.8 156
23.1
173
13.5
High school
957
48.8
319
47.4
637
49.6
College diploma or higher
673
34.4
199
29.5
475
36.9
Average years of school
12.35
11.98
12.54
Sex
Male
1,235
63.0
511
75.8
724
56.4
Female
724
37.0
163
24.2
561
43.6
Spouse
Yes
1,509
77.1
575
85.3
935
72.7
No
450
22.9
99
14.7
351
27.3
No. of family 
members
Average (people)
3.51
3.61
3.46
Householder
Yes
1,205
61.5
518
76.8
687
53.5
No
754
38.5
156
23.2
598
46.5
Economic 
activity 
participation 
status
Regular worker 
253
12.9
92
13.7
161
12.5
Temporary worker
165
8.4
29
4.2
136
10.6
Daily laborer
238
12.2
62
9.2
176
13.7
Self-employed
695
35.5
406
60.3
288
22.4
Others
608
31.1
85
12.7
523
40.7
Disposable 
income
a
(W 10,000)
≤ 1,000
291
14.8
73
10.8
218
16.9
1,000~2,000
823
42.0
248
36.8
575
44.7
2,000~3,000
526
26.8
185
27.5
340
26.5
3,000~4,000
173
8.8
77
11.5
95
7.4
≥ 4,000
148
7.5
90
13.3
58
4.5
Average
2,178.7
2,547.9
1,985.2
SOURCE: Author, based on data from the Korea Welfare Panel Study 2008.
A. Adjusted for family size using √n. W 1,125 ≈ $1.

168 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
for Agriculture and Fishery Structure Adjustment. People who are entitled to the subsidy 
work in the agriculture, forestry, livestock, and fishery businesses and are individually 
insured (or voluntarily and continuously insured). The subsidy is scheduled to be termi-
nated by the end of 2014.
Initially, farmers and fishers were assisted with a third of the lowest contribution 
amount; later the amount gradually increased. Currently, they are given half of the total 
contribution, up to an income ceiling; when income exceeds the ceiling, a fixed amount 
is applied (table 8.6).
If the matching subsidy has positive effects on increasing the compliance rate of 
farmers and fishers, the expansion of such a subsidy program to other noncompliant 
groups may be an effective policy instrument that could reduce blind spots in national 
TABLE 8.5  Probit regression estimations of determinants of compliance with Korea’s national 
pension plan among individually insured people
Dependent variable
Individually insured or noninsured
Model 1
Model 2
dy/dx
Standard 
error
dy/dx
Standard 
error
Male
0.0758***
(0.028)
0.0736***
(0.028)
Age
0.0375**
(0.016)
0.0355**
(0.016)
Age 2
−0.0002
(0.000)
−0.0002
(0.000)
Economic activity 
participation status
(Basis: temporary 
workers)
Regular workers 
0.1835***
(0.063)
0.1822***
(0.063)
Daily laborers
0.0023
(0.058)
0.0048
(0.058)
Self-employed
0.3129***
(0.052)
0.3150***
(0.052)
Other
a
−0.0513
(0.052)
−0.0492
(0.052)
Education
(Basis: middle school or 
below)
High school
−0.0475
(0.037)
College or higher
−0.0399
(0.042)
Years of schooling
−0.0146
(0.021)
ln (disposable income)
b
0.1144***
(0.023)
0.1140***
(0.023)
Number of observations 
1,933
1,933
Log pseudo-likelihood
−979.729
−980.054
Pseudo R
2
0.2127
0.2124
NOTE: ***p < 0.01, **p < 0.05, *p < 0.1.
A. Includes unpaid family workers, unemployed, and economically nonactive population. 
B. Adjusted for family size using √n.

8.  MATCHING CONTRIBUTIONS AND COMPLIANCE IN KOREA’S NATIONAL PENSION PROGRAM 
169
pension coverage. Differences in compliance between the group of farmers and fishers 
who are individually insured and receive the subsidy and other self-employed groups that 
are individually insured but do not receive a subsidy are therefore studied.
CONCEPTUAL FRAMEWORK
Intuitively, a contribution subsidy should provide incentives for pension compliance, 
especially to workers who are not currently making contributions. The effect of the 
subsidy can be divided into two parts. First, it can help reduce delinquency among 
people who are already individually insured but are not able to make their contributions 
because of economic difficulties. The current contribution rate of 9 percent of income 
can be burdensome to low-income households. Reducing this contribution to 4.5 per-
cent through a matching subsidy would reduce liquidity constraints (the “contribution 
effect”).
Second, a substantial portion of individually insured people who earn income decide 
to remain exempt by concealing their income from the national pension system. Provid-
ing a contribution subsidy to these groups would make it more advantageous to them 
to register with the scheme by reporting their income to the authorities. The contribu-
tion subsidy would provide incentives for voluntary income reporting, reducing the blind 
spots in pension coverage. The magnitude of this “extensive margin” will depend partly on 
the generosity of the benefits provided.
The contribution subsidy could also encourage people in the targeted groups to 
report their income levels more honestly. Underreporting of income levels, with short-
sighted intentions to reduce current contribution burdens, is believed to be prevalent 
among individually insured people. It is possible because of the government’s limited abil-
ity to gather accurate information on personal income. A contribution subsidy might 
induce people to report their income more truthfully by reducing the economic burden of 
doing so. This “intensive margin” will be greater if the contribution subsidy increases with 
the level of reported income.
A natural experiment can be exploited by comparing the behavior of individually 
insured people who received the matching subsidy with that of individually insured peo-
ple who did not. A few strong assumptions are necessary—namely, that all individually 
insured people are ex ante homogeneous in compliance, regardless of the type of work 
they do, and that non-observables (for example, differences between the groups in “choice 
architecture,” such as ease of contribution payment) are not relevant.
TABLE 8.6  Size of matching subsidy to farmers and fishers in Korea
Standard monthly income
Size of subsidy
Less than W 790,000 
4.5% of income (half of total contribution)
More than W 790,000
W 35,550 (about $32)
SOURCE: http://english.nps.or.kr. 
NOTE: W 790,000 (about $700) corresponds to the 18th of the 45 grades in the standard monthly income table. 

170 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
DATA
Empirical analysis of the effects of the contribution subsidy requires detailed information 
on individual characteristics and compliance behaviors of various types of individually 
insured people, including farmers and fishers. The Korea Welfare Panel Study is the only 
source of data that contains detailed information on compliance with the national pen-
sion and on farmers and fishers. This study conducted a quantitative analysis using these 
data. The sample included 704 self-employed workers age 18–59 who were individually 
insured by the national pension. Among these workers were 142 farmers and fishers who 
received the contribution subsidy in 2007.
The narrow coverage of the national pension is indeed a serious problem among 
the self-employed, with 40 percent not contributing (table 8.7). The average delinquency 
rate of farmers and fishers was 9.2 percent—far lower than the 13.9 percent of other 
self-employed workers. The proportion of exempt payers was also significantly lower 
(18.3 percent for farmers and fishers versus 29.9 percent for other self-employed workers). 
The rate of contributors among farmers and fishers was 16.3 percentage points higher 
than that of other self-employed workers.
TABLE 8.7  Compliance with Korea’s national pension by subsidy recipients and nonrecipients, 
2007
Category
Total
Farmers and fishers
Other self-employed
Contributors
419 (59.5)
103 (72.5)
316 (56.2)
Noncontributors
285 (40.5)
39 (27.5)
246 (43.8)
Delinquents
91 (12.9)
13 (9.2)
78 (13.9)
Exempt payers
194 (27.6)
26 (18.3)
168 (29.9)
Total
704(100.0)
142(100.0)
562(100.0)
SOURCE: Author, based on data from the 2008 Korea Welfare Panel Study.
NOTE: Figures in parentheses are percentages of the total.
Table 8.8 compares the major characteristics of farmers and fishers with those of 
other self-employed workers. The average age of farmers and fishers surveyed was 50.9, 
considerably higher than the average age of other self-employed people (44.5). More than 
half of farmers and fishers were 50 or older.
4
 The average disposable household income of 
farmers and fishers was about three-quarters that of other self-employed workers. These 
data imply that, on average, rural households were more likely than other households to 
face binding liquidity constraints in paying pension contributions and in saving for post-
retirement consumption. The average education level of farmers and fishers was signifi-
cantly lower than the average level of other self-employed people, and the percentage of 
men was higher.

8.  MATCHING CONTRIBUTIONS AND COMPLIANCE IN KOREA’S NATIONAL PENSION PROGRAM 
171
TABLE 8.8  Demographic characteristics of recipients and nonrecipients of subsidized contribution 
to Korea’s national pension plan, 2007
Category
Total
Farmers and 
fishers
Other 
self-employed
No. of 
samples
%
No. of 
samples
%
No. of 
samples
%
Total
704
100.0
142
100.0
562
100.0
Age
18–30
19
2.7
0
0.0
19
3.4
31–40
170
24.2
9
6.3
161
28.7
41–50
296
42.1
52
36.6
244
43.4
51–59
219
31.1
81
57.0
138
24.6
Average (years)
45.8
50.9
44.5
Educational 
Background
Middle school or lower
177
25.1
76
53.5
101
18.0
High school
332
47.2
59
41.6
273
48.6
College diploma or 
higher
195
27.7
7
4.9
188
33.5
Average years of 
schooling
11.8
9.6
12.4
Sex
Male
582
82.7
128
90.1
454
80.8
Female
122
17.3
14
9.9
108
19.2
Spouse
Yes
603
85.6
128
90.1
475
84.5
No
101
14.4
14
9.9
87
15.5
Family size
Average (people)
3.7
3.9
3.6
Householder
Yes
588
83.5
129
90.9
459
81.7
No
116
16.5
13
9.2
103
18.3
Disposable 
income
a
(W 10,000)
≤ 1,000
113
16.1
39
27.5
74
13.2
1,000~2,000
284
40.3
62
43.7
222
39.5
2,000~3,000
181
25.7
26
18.3
155
27.6
3,000~4,000
57
8.1
6
4.2
51
9.1
≥ 4,000
69
9.8
9
6.3
60
10.7
Average
2,311.0
1,842.6
2,429.4
SOURCE: Author, based on data from the 2008 Korea Welfare Panel Study.
A. Adjusted for family size using √n. W 1,125 ≈ $1. 

172 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
EMPIRICAL RESULTS
This study uses a probit regression model that includes dummies for farmers and fish-
ers. The model estimates differences between the group of individually insured farmers 
and fishers and other self-employed groups. The effect of the contribution subsidy can 
be decomposed into the contribution effect and the registration effect, each of which is 
estimated separately.
Table 8.9 summarizes the estimation results for the entire sample of self-employed 
people. The first column reports the result when the probability of contribution payment 
is estimated for income reporters. The result reveals that among income reporters, dispos-
able income is the most important variable affecting the decision to make a pension con-
tribution. Other variables, such as age and education, have some explanatory power, but 
their statistical significance is relatively weak. The coefficient of the dummy variable for 
farmers and fishers is positive and statistically significant within the confidence interval 
of 90 percent. After controlling for individual characteristics, the probability of making a 
pension contribution was 7.4 percentage points higher among the subsidized farmers and 
fishers than it was among other nonsubsidized. The contribution subsidy thus had a non-
negligible effect on reducing delinquency among the subsidized group by lightening the 
financial burden of participating.
5
TABLE 8.9  Contribution, registration, and total effects of subsidy on pension behavior of farmers 
and fishers in Korea
PROBIT REGRESSION RESULTS
Dependent variable
Contribution effect
Registration effect
Aggregate effect
dy/dx
Standard 
error
dy/dx
Standard 
error
dy/dx
Standard 
error
Male dummy
−0.0309
0.041
0.0565
0.047
0.0278
0.051
Schooling years
0.0110*
0.007
−0.0082
0.007
0.0004
0.007
Age
−0.0442*
0.025
0.0259
0.023
−0.0097
0.027
Age2
0.0006**
0.000
−0.0002
0.000
0.0003
0.000
ln(Disposable income)a
0.0677**
0.027
0.0696***
0.027
0.1088***
0.031
agri_dummy
0.0742*
0.038
0.0669
0.045
0.1108**
0.051
Number of observations
501
691
691
Log pseudo-likelihood
−218.104
−387.327
−435.833
Pseudo R
2
0.0694
0.0469
0.0649
SOURCE: Author, based on data from the 2008 Korea Welfare Panel Study.
NOTE: ***p < 0.01, **p < 0.05, *p < 0.1. The contribution effect (the probability of contribution payment) is estimated for 
income reporters. The registration effect (the probability of income reporting) is estimated for the entire sample. The aggre-
gate effect is estimated for the entire sample, including people who are exempt. Adjusted for family size using √n.

8.  MATCHING CONTRIBUTIONS AND COMPLIANCE IN KOREA’S NATIONAL PENSION PROGRAM 
173
The estimates of the registration effect are reported in the second column of the 
table. When the probability of income reporting is estimated for the entire sample, the 
coefficient of the dummy variable for the subsidized group is positive, as expected, imply-
ing that the subsidized group is 6–7 percentage points more likely to report income to the 
national pension system than the nonsubsidized group. The estimate is not statistically 
significant, however.
The third column in the table reports the estimation results of the overall effect 
of the contribution subsidy when the probability of contribution payment is estimated 
for the entire sample, including exempt payers. The estimated coefficient of the subsidy 
dummy is greater in absolute value and is statistically significant: the estimated probability 
of paying a contribution was 11 percentage points higher for farmers and fishers and sta-
tistically significant at the 95 percent level. The empirical evidence on the positive effect 
of the contribution subsidy seems to be stronger in this case.
Differences in compliance are illustrated in figure 8.2. Compliance rates of the sub-
sidized farmers and fishers are higher than those of the nonsubsidized group across all 
income levels. The average compliance gap between the two groups is more than 16 per-
centage points. 
The estimation results imply that the contribution subsidy plays a significant role 
in encouraging participation of farmers and fishers. One would expect the subsidy to 
FIGURE 8.2  Payment of pension contributions by subsidy recipients and nonrecipients in Korea, by 
level of income, 2007 
64%
73% 
73% 
83%
100%
73%
42% 
55% 
58% 
69% 
63%
56%
50% 
59% 
60%
70%
68%
60%
0  
10  
20  
30  
40  
50  
60  
70  
80  
90  
100  
W 1,000
or lower 
W 1,000~
W 2,000 
W 2,000~
W 3,000 
W 3,000~
W 4,000 
W4,000
or higher
average
(W 10,000)
farmers/fishers
other self-employed 
total