income inequality in korea, 1933-2010: evidence from
TRANSCRIPT
1
Income Inequality in Korea, 1933-2010:
Evidence from Income Tax Statistics
Nak Nyeon Kim and Jongil Kim
Abstract
This paper constructs long-term series of top income shares and top wage income shares using income
tax statistics to study the evolution of income inequality in Korea. Income concentration was very
high before WWII when Korea was under colonial rule but dropped sharply after WWII. It remained
low during the period of rapid industrialization but has ascended since the mid-1990s. This U-shaped
pattern of top income shares in Korea is similar to those in other industrialized countries but the
factors for the shape may be different. We attribute the drastic fall in income concentration after
WWII to economic collapse and political disruption from regime change after liberalization. We
argue rising income concentration last 15 years could be attributable to two factors. First, we point out
the drastic transition of major industries toward technology-intensive sectors along with deepening
economic relationship with low-wage countries such as China caused diverging demands for different
types of workers. We find the average wage income of the bottom income groups even declined in
real terms last 15 years. Second, we suggest two institutional factors such as falling marginal tax rate
since the 1980s and post-crisis changes in corporate governance system enforcing competition among
the top management through performance-based rewards played an important role in increasing top
incomes.
I. Introduction
The indices of income inequality such as Gini coefficient and poverty rate are constructed based on
Household Income and Expenditure Survey by the Korea Statistical Office.1 Although the household
survey is a suitable source to compute the indices of income inequality, it does not allow us to study
the long-run evolution of income inequality in Korea. The raw data of the household survey are
available since 1990, the earliest year the official statistics of income inequality is reported by the
Korea Statistical Office.2 The household survey did not cover single-person households before 2006.
In addition, it provides household income only for wage workers in the earlier years by excluding
income of self-employed and family workers. In addition, it was found that the household survey
leaves out a considerable number of top income households in the sample and severely understates
household financial income(Kim and Kim, 2013). Due to this problem of under-coverage and under-
reporting, the existing indices of income inequality do not allow us to study the evolution of income
equality in Korea because a homogeneous series is available only for a small number of recent years
and even available series may not adequately reflect real situation of income inequality.
Instead of using the indices based on the household survey, this paper constructs long-term series of
income concentration to study the evolution of income inequality in Korea. We compute income
shares and wage income shares for top income groups such as the top 1% of income distribution by
using income tax statistics. Although top income shares do not provide information on the income
1 To be exact, income inequality indices such as Gini Coefficient and income decile distribution are computed
based on Household Income and Expenditure Survey by the Korea Statistical office and Farm household
Economy Survey by the Ministry of Agriculture and Forestry. 2 The Gini coefficients, decile income distributions, and poverty rates since 1990 are reported in the website
(http://kostat.go.kr) of the Korea Statistical Office.
2
distribution of lower income groups, it allows us to construct a homogeneous long-term indicator of
income inequality. It also let us study the evolution of income inequality in Korea in comparative
perspective since top income shares have been constructed for many countries such as France(Piketty,
2003), the U.S.(Piketty and Saez, 2003), the U.K.(Atkinson, 2005), and Japan(Moriguchi and Saez,
2008).3
The income tax statistics in Korea goes back to the colonial period. The composite income tax
which imposes tax on the sum of various types of income was introduced in 1934 for the first time.
Since then, there are three major changes in tax system until now in addition to numerous minor
revisions and thus available information from income tax statistics varies over time. Notwithstanding
missing periods due to limited information from the tabulations of income tax statistics, this paper
investigates the long-term trend of income inequality in Korea from 1933 to 2010.
During this period, Korea underwent several different regimes. From 1933 to 1945, Korea was
under Japanese colonial rule and experienced a wartime control system after 1937. After liberalization
in 1945, the Korean Peninsula was divided into two nations in the South and the North. The Korean
War broke out in 1950 and continued three years. After the war, South Korea achieved long and rapid
economic growth although its growth rate subsided after the economic crisis in 1997. This paper will
provide a starting point to discuss how income inequality has evolved under these different regimes.
The paper is organized as follows. Section II describes the data and outlines the estimation methods.
Section III presents our findings from the construction of top income shares and investigates the
causes of the observed changes in trend. In Section IV, we present the trend of top wage income
shares and explore the causes in comparison with other countries, particularly the U.S. and Japan.
Section V concludes. The detail description of our data and methods, as well as a complete set of
results, will be presented in Appendix.
II. Data and Methodology
1. Construction of Top Income Shares Series
Top income share is computed by dividing income accruing to a specified top income group by
total income. For this, we have to define income that should be counted and individuals belonging to
specified top income groups such as the top 1%.
Our primary data source is income tax statistics published in the Annual Statistical Report on
National Tax (Kuksae Tonggae Yeonbo, the annual report from now on) by the National Tax
Administration of Korea and the Japanese Government General of Korea. The annual report provides
tables with the number of taxpayers, the amount of reported income, and the amount of income tax by
income brackets.
We define income as a gross income before deductions of income and payroll taxes paid by
individuals, but after employers’ payroll taxes and corporate income taxes. It includes all types of
income such as capital income (interest and dividends), rental income (rents from real-estate lease),
unincorporated business income (including self-employment income), and wage income (salaries,
wages, and bonuses), pension, and other small items. However, realized capital gains are not included
in income in our definition. Capital gains from land and house have been partially taxed due to special
treatments and exemptions and those from listed stocks have not been taxed except for large
shareholders designated by law in Korea. In addition, realized capital gains are volatile and lumpy as
opposed to a steady source of annual income.
3
Top income shares series of various countries can be obtained from the World Top Income
Database(http://topincomes.g-mond.parisschoolofeconomics.eu). Top income shares series seem to become
an alternative measure of income inequality as they are accumulated. OECD report on income inequality paid
attention to top income shares along with traditional indices of income inequality(OECD, 2011). So far,
Korea has been left in the blank in studying top income shares. We expect our study will fill in the blank for
Korea.
3
Top income groups are defined relative to the total number of adults, defined as 20 years old and
above. The total adult population can be computed directly from official population statistics provided
by the Korea Statistical Office since 1960. For the period earlier than 1960 when population by age
groups is available only for the years of population census, we interpolate data in missing years by
using the growth rate between two census years.
Other candidates for the control total for population could be total number of taxpayers or total
employment. Because of tax exemption and withholding tax system, only a small fraction of
individuals file income tax returns. Therefore, the number of total taxpayers is not a consistence
reference for time-series and cross-section comparison since its scope varies among different tax
systems. Total employment is too narrow since it excludes individuals relying on non-labor sources
such as pension and capital income. Although the adult population, our control total, includes
economically inactive population, it provides a relatively proper reference for comparison over time
and across countries. Our definition is same as that of Moriguchi and Saez(2008) who constructed top
income shares series for Japan.
We cannot compute total income from income tax statistics because the tabulations in income tax
statistics do not count taxpayers who do not file income tax returns. Thus, we rely on National
Accounts to compute the control total for income. We obtain the amount of total income after
subtracting items which do not belong to personal earning from the table of income by institutions in
National Accounts. We subtract employers’ social contribution, imputed rents to owner-occupier, and
financial intermediation services indirectly measured(FISIM) from compensations to employees,
operating surplus, and interest in financial income, respectively. For the colonial period when the
statistics of income by institutions is not available, we obtain total income by subtracting imputed
rents from factor income which is the sum of compensation to employees and operating surplus. This
method cannot exclude retained corporate income but it is an ignorable amount at that time.
To compute income accruing to top income groups (in this study, the top 0.01%, 0.05%, 0.1%,
0.5%, 1%, 5%, and 10%), we use method used by Piketty and Saez(2003). The top tail of income
distribution is well approximated by a Pareto distribution. We estimate Pareto coefficients bracket by
bracket for each year using the distribution tables of income tax statistics in the annual report. We
employ a parametric interpolation method to estimate threshold and average income levels for top
income groups. We then obtain total income accruing to the respective top income groups simply by
aggregating income above the threshold. Finally, the income shares of respective income groups are
computed by dividing the income of each group by the control total for income.
Table 1 presents estimated threshold and average income levels for top income groups in 2010. The
size of the top 0.01% income group is 3,895 and one should earn more than 1.10 billion won
(US$ 995,346) to belong to this group. Its average income is 2.73 billion won.4 Top 5-1% denoted in
the table is the top 5% income group excluding the top 1% income earners. The size of this group is
1.55 million and its average income is 72.5 million won (US$ 62,747).
<Table 1> Threshold and Average Income Levels for Top Income Groups in 2010
Notes: (1) Unit of income is thousand won.
4 Annual average market exchange rate in 2010 (1,156 Korean won per one US dollar) was used for conversion.
Percentile
Threshold
Threshold
income levelsIncome groups
Number of
adults
Average income
levels
Full Population 38,946,431 16,383
Top 10% 36,201 Top 10-5% 1,947,322 46,724
Top 5% 57,063 Top 5-1% 1,557,857 72,535
Top 1% 104,948 Top 1-0.5% 194,732 113,443
Top 0.5% 135,564 Top 0.5- 0.1% 155,786 173,557
Top 0.1% 283,886 Top 0.1-0.01% 35,052 467,515
Top 0.01% 1,104,380 Top 0.01% 3,895 2,730,837
4
(2) Top 10-5% is the top 10% excluding the top 5%. It is same for others.
For our purpose, we basically use information on self-assessed income in the tabulation of
composite income tax. Self-assessed income filed for composite income tax returns includes various
types of personal income such as interest, dividends, rents, and pension as well as wage and business
income. However, since the tabulation of income tax in the annual report changed along with tax
reform, we cannot construct top income shares for some years. Figure 1 displays major changes in tax
system since colonial period. For the period of 1933-1953, household’s total income (the class III
income following the terminology at that time, which is close to the current definition of composite
income) was taxed if a household, the tax unit at that time, earns total income more than exemption
point. For the period of 1954-1975, earnings from different sources were taxed separately because
Korea did not adopt composite income tax system. Therefore, it is not possible to compute top income
shares for this period since the tabulation in the annual report does not allow us to extract required
information.
<Figure 1> Evolution of Income Tax System in Korea
Source: Income Tax Law of Korea obtained from the website of the Ministry of Government
Legislation(http://www.law.go.kr).
Notes: (1) Periods specified in the parentheses are those during which the tax statistics are available.
(2) Income tax statistics used in this paper are shaded.
The composite income tax system was introduced in 1975. However, due to extensive withholding
system, there are years when financial income is not counted in self-assessed income for composite
income tax returns. For the period of 1975-1995, interest and dividends were withheld at source and
excluded from composite income tax returns. Since 1996, these financial income items were included
if the total amount of financial income exceeds a certain level(currently 40 million won per person).5
Due to data limitations, we cannot adjust for this inconsistency in counting financial income among
periods. Similarly, for the period of 1933-1953, part of interest income was excluded. Interest income
from some bonds specified by law (so called the class Ⅱ income) was withheld at source.
Next, a fraction of top income groups could be exempted from filing composite income tax returns.
Although the composite income tax system was introduced in 1975, individuals only with wage or
pension income are not required to file composite income tax returns if their earnings from financial
assets including interest and dividends do not exceed a certain amount, currently 40 million won.
Thus, income accruing to this group of top income earners is not counted in the tables of composite
income tax returns. So, we combine the tabulations of composite income tax and wage income tax to
make up for these missing top wage earners. We then remove double-counted income of overlapping
taxpayers, wage earners having multiple income sources, who are included in both tabulations of the
composite income tax and wage income tax. However, we cannot make this adjustment from 1975 to
1994 since the annual report does not provide the statistics of wage income tax withheld at source by
5 This practice was suspended during 1998-2000 due to economic crisis.
1933-1953 1954-1975 1976-1995 1996-present
Composite
income taxation
Class Ⅲ income tax (1933-
1942)
Class Ⅱ income tax(interest)
Business income and rents
Tax unit Household Individual
Composite income (1976-present)
Separate income
taxation
Interest (Financial income included
in composite income)Dividends
Wage income (1957-1974, 1979-1985, 1995-present)
Others
5
income brackets. Thus, we cannot construct top income shares combining the two income tax
statistics during this period. Although we filled some missing years from 1979 to 1985 using internal
source of the National Tax Administration, top income shares could not be estimated for remaining
years.6
In addition, one may concern about the possibility of tax evasion. Tax evasion will not be
significant for the types of income, tax for which is withheld at source but may be so in case of
business income which is self-assessed by taxpayers for tax returns.7 We don’t have reliable
information on tax evasion and thus cannot make any adjustment for this.
Finally, for the period of 1933-53, the unit of income tax was family. For this period, tax was
imposed on the aggregate income earned by all family members in the household and was allotted to
individual family members in proportion to their income levels. The annual report for this period
provides two different kinds of tabulation on income tax statistics, one in terms of the number of tax
units by (household) income brackets and the other in terms of the number of individuals by tax-paid
brackets. The tabulation of individuals by tax-paid brackets can be converted to that of individuals by
income brackets. Here, we use this converted tabulation of individuals by income brackets for this
period, which is consistent with what we use for estimation in later periods.
2. Construction of Top Wage Shares
We use the same methodology to compute top wage shares. We use published tables classifying tax
returns by size of salaries and wages from the annual report. Top wage shares can be constructed
longer than top income shares since wage income tax statistics by income brackets go back to the
1950s. The share of top wage income is computed by dividing wage income accruing to top wage
groups by total wage income.
Here, top wage groups are defined relative to the total number of employment. Therefore, our
definition of wage earners is consistent with the definition of employment in official statistics which
includes both regular and temporary workers including daily workers. It is obtained from the Annual
Survey on Economically Active Population (ASEAP).
Because the tables in the annual report exclude a large proportion of low wage earners under the
exemption point, we cannot use the total sum of wage income in the annual report except for recent
years when the relevant information is provided. Therefore, again we use the wage and salaries in
National Accounts to compute the control total for wage income. It allows us to compute top wage
income shares back to 1963, the earliest year for which total employment is available in the ASEAP.
Thus, wage income in our definition includes wages, salaries, bonuses, allowances and taxable part of
non-cash compensation, but excludes retirement benefits.
However, wage and salaries in National Accounts include pay to the workers who are not officially
classified as employed in the ASEAP. Therefore, we have to remove wage income of this group from
total wage income in National Accounts to be consistent with our definition of wage earners. The
annual report in 2009 provides the number of workers by detailed types such as wage earners, both
taxed and exempted, and daily workers whose income is withheld at source. When we aggregate all
these types of workers reported in the annual report, we find the aggregate exceeds the number of
6 Since 2009, the annual report provides income tax withheld at source by income brackets for all the items
such as pension, wages for daily workers, and part of business income which is exempted from filing
composite income tax returns. It allows us to estimate top income shares including other income items in
addition to wage income. However, we found it does not make any significant difference. It is due to the fact
that the income levels of most individuals with income tax withheld at source for these items are not so high
to be included in top income groups. Thus, our estimate combining composite income tax statistics and wage
income tax statistics seems to cover almost all the individuals with high taxable incomes to be included in top
income groups since most top income earners file either composite income tax or wage income tax returns. 7 When we compare National Accounts and the annual report, we found that about 26% of business income
seems to evade tax. Other types of income such as interest, dividends, and wage, for which tax are usually
withheld at source seem well detected by income tax statistics. See Kim and Kim(2013).
6
total employment in the ASEAP. It indicates there are a considerable wage income earners not defined
as employed in the ASEAP, our control total for workers. Therefore, we make the number of workers
in the annual report identical with that in the ASEAP by deducting daily workers from the lowest
wage level who tend to be not classified as employed by the ASEAP. Then, we subtract wage income
of these deducted daily workers from total wage income in National Accounts and obtain our control
total for wage income. We find total wage income in National Accounts is greater than our control
total for wage income by 6.7% in 2009. However, since the annual report does not provide detailed
statistics for other years, we cannot do this exercise for control total for wage income. Therefore, we
deflate wage and salaries in the National Account each year by the same rate of 6.7% computed for
2009.
III. Top Income Shares in Korea, 1933-42, 1976-2010
1. Historical Background
Figure 2 displays the evolution of per capita GDP of Korea last 100 years. First, during the pre-
World War II(WWII) period when Korea was under colonial rule, per capita GDP of Korea grew
modestly as fast as those of Japan or the U.S. and stayed about 20 percent of the U.S. level. Second,
the average income declined sharply due to the disruption from WWII, the division of country in 1945,
and the Korean War in 1950-53. The average income of Korea in the 1950s plummeted to the level
less than 10 percent of the U.S. level. Third, in the mid-1960s, the Korean economy began to grow
rapidly and now Korea’s average income approaches to about 70 percent of the U.S. level. During this
period, Korea achieved rapid industrialization in a more compressed way than Japan which also
caught up with the U.S. in a short time span relative to other developed countries.
<Figure 2> GDP per capita in United States, Japan, and South Korea
Source: Korea’s GDP per capita from Kim ed.(2012) and those of other countries from
Maddison(http://www.ggdc.net/maddison/maddison-project).
Note: GDP per capita in 1990 international Geary-Khamis dollars(log scale in the left axis) and
the ratio of GDP per capita of Korea with respect to the U.S. (percentage in the right axis)
Figure 3 shows long-run trends of the growth rates of population and total GDP and the level of
consumer price index(CPI) in Korea. Due to political tumult after liberation and the collapse of
economy during the War, Korea experienced hyperinflation as shown by a sharp rise of the CPI during
this period. It made a big impact on the distribution of income. Before the Korean War, South Korea
witnessed a big surge in population as Koreans who went abroad during colonial period returned
0
25
50
75
100
125
150
100
1,000
10,000
100,000
18
70
18
75
18
80
18
85
18
90
18
95
19
00
19
05
19
10
19
15
19
20
19
25
19
30
19
35
19
40
19
45
19
50
19
55
19
60
19
65
19
70
19
75
19
80
19
85
19
90
19
95
20
00
20
05
20
10
USA
Japan
South Korea
Korea/USA
7
home and refugees escaped to the south from communist regime. Population decreased sharply as
many people died during the Korean War. After the War, Korea experienced a baby boom which
increased population growth rate temporarily. These factors may explain slow recovery of average
income in Korea after liberalization.
Figure 3 also presents the trend of GDP growth rate over time. Under Japanese colonial rule, Korea
experienced a modest growth with relatively high fluctuation. The fluctuation was mainly caused by
changing crop situation since agriculture was the major industry in Korea at that time. In the post-
WWII period after recovery from politico-social chaos, Korea sustained a long span of high growth.
Korea experienced negative growth just for two years in 1980 and 1998 during this long span of
economic growth. After 30 years of high growth, the GDP growth subsided gradually as economic
development of Korea enters a mature stage. Last 10 years, Korea grew less than 5 percent per annum
which is relatively low compared to high growth rates in previous 30 years.
<Figure 3> Population and GDP Growth and Consumer Price Index in Korea
Sources: Korea Statistical Office(KOSIS); Kim ed.(2012); Park and Kim(2011).
Notes: 1) The growth rates of population and GDP are 3-year moving averages of annual growth rates.
2) The growth rates of population and GDP are in percentage in the left axis and the consumer price
index (2010=100) is in the right axis.
2. Trends in Top Income Shares
Figure 4 displays three series of the top 1% income share; series 1 constructed from the composite
income tax statistics, series 2 from wage income tax statistics8, and finally, series 3 constructed after
combining the composite income tax and wage income tax statistics9. The series 3 is most
comprehensive in coverage among three series and could be considered as our final estimate of top
income shares. The difference between series 3, combining two income tax statistics and series 1
based on composite income tax statistics is exactly income share of the top 1% income group, with
wage income only, who do not file composite income tax returns.10
We cannot directly compare the
8 The trend of wage income concentration will be discussed in detail in the next section.
9 For the pre-WWⅡ period this procedure is not needed, since the composite income tax covers all the wage
income above exemption point. 10
If we construct income shares of smaller fractiles such as the top 0.1% or 0.01% within the top 1%, the
difference between the two estimates of top income shares decreases and become insignificant. The reason is
there are few who do not file composite income tax returns among the upper income earners within the top
1%.
0.000001
0.00001
0.0001
0.001
0.01
0.1
1
10
100
-15
-10
-5
0
5
10
15
20
25
19
11
19
15
19
19
19
23
19
27
19
31
19
35
19
39
19
43
19
47
19
51
19
55
19
59
19
63
19
67
19
71
19
75
19
79
19
83
19
87
19
91
19
95
19
99
20
03
20
07
GDP
population
CPI
8
levels of top wage income shares (series 2) and other two top income shares (series 1 and 3) due to
different definitions of control total. Though, the trends of top income shares series do not look much
different from that of top wage shares series during the periods when both series are constructed.11
Therefore, we may conclude that income concentration in Korea, very high during the pre-WWII
period, dropped sharply after WWII, stayed stable at low level during the period of high growth, and
increased noticeably after the mid-1990s.12
.
<Figure 4> Top 1% income shares in Korea, 1933-2010
Source: Appendix.
Note: Top income shares series denoted as series 1 is constructed using composite income tax
statistics. Series 2 is constructed using the wage income tax statistics. Series 3 is estimated
by combining tax statistics of composite income tax and wage income tax.
Figure 5 plots income shares of the top 1% and the next 4% (top 5-1%) based on the combined
income tax statistics. The sharp drop of the top 1% income share after WWII is contrasted with the
relative stability of the top 5-1% income share which shows no significant change after the War. It
suggests that the sharp decline of income concentration during this period could be attributable to the
top tail of income distribution within the top 1 %.
Figure 6 investigates this point further by decomposing the top percentile into three subgroups. It
shows income shares of the top 0.1%, the next 0.4% (top 0.5-0.1%), and the bottom half of the top 1%
(top 1-0.5%). As the income share of the top 1% fell after WWII, all three subgroups exhibit similarly
falling income shares but they are distinguished from one another in terms of magnitude. The higher
income fractile within the top 1% experienced the larger fall of income share after WWII. In the pre-
WWII period, the top 0.1% took a larger share of income than lower income fractiles even though it
included 4 or 5 times less population. After WWII, the income share of the top 0.1% fell drastically
from 7% to 2% and became lowest among the three subgroups.
With the rising top 1% income share since the mid-1990s, a similar pattern is observed in the
opposition direction among the subgroups. All three subgroups have been rising in income shares
since the mid-1990s but the higher income fractiles exhibited the more rapid rise. The top 0.1%
11
Wage income accounts for more than half of total income. 12
One should note that both estimates of top income shares (series 1 and 3) were not adjusted for the
inconsistent coverage of financial income over time. As already told, financial income over a certain amount
began to be taxed as composite income in 1996 but this practice was suspended for 3 years (1998-2000) after
economic crisis. Therefore, our estimates of top income shares before 1995 and during 1998-2000 could be
underestimated because financial income was not counted for these periods. This factor partly explains a big
drop of top income shares in 1998.
-
2.00
4.00
6.00
8.00
10.00
12.00
14.00
16.00
18.00
20.00
19
33
19
36
19
39
19
42
19
45
19
48
19
51
19
54
19
57
19
60
19
63
19
66
19
69
19
72
19
75
19
78
19
81
19
84
19
87
19
90
19
93
19
96
19
99
20
02
20
05
20
08
Series 3
Series 1
Series 2
9
income share which stayed lowest in 1995 rose to the highest level among the subgroups in 2010. It
indicates that deepening income concentration since the mid-1990s accompanied the sharper rise of
income of the upper income tails in the top percentile.
<Figure 5> Top 1% and Next 4% Income Shares in Korea, 1933-2010
Source: Appendix.
Note: Top 5-1% denotes the income share of the top 5% excluding the top 1 %.
<Figure 6> Decomposition of Top 1% Income Shares in Korea, 1933-2010
Source: Appendix.
Note: Top 0.5-01% denotes the income share of the top 0.5% excluding the top 0.1 %. Top 1-
0.5% denotes the income share of the top 1% excluding the top 0.5 %.
Figure 7 displays top 1% income share series in Korea with those in other countries to provide a
comparative perspective. Korea is not so much different from other countries in overall pattern of
trend by showing a U-shaped pattern of income concentration. The top 1% income share in Korea was
as high as those in other countries in the pre-WWII period.
Notwithstanding missing years, we may guess that Korea’s top income share fell sharply as those of
other countries after WWII. The top income share in Korea remained low until the mid-1990s, while
the U.S and the U.K exhibited a rising top income shares starting from the 1980s.13
In the mid-1990s
13
Our guess is based on figure 4 which shows a complementary series based on wage income tax statistics
which goes back to 1963.
0
5
10
15
20
2519
33
1936
1939
1942
1945
1948
1951
1954
1957
1960
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
top 1%
top 5-1%
10
when the top income share in Korea began to rise, the level of top income shares in Korea was similar
to those in Japan and France which show a different pattern from the U.S. and the U.K. However,
income concentration in Korea since the mid-1990s has ascended much faster than in Japan and
France. In terms of slope, rising trend in Korea is similar to those in the U.S and the U.K. As a result,
the top income share in Korea in 2010 was located in between two groups of countries which show
diverging patterns in income concentration.
<Figure 7> Top 1% Income Shares in Korea and Selected Countries
Source: Appendix 1; The World Top Income Database
To better understand the mechanism that led the shift of trend in income concentration in Korea, we
compute the top income composition for selected years. For 1933-42 in pre-WWII period, the annual
report allows us to compute income composition of total taxpayers in the tabulation of class III
income tax. Thus, the number of total taxpayers in the annual report accounts for a varying fraction of
the top income group from the top 1.4% in 1933 to the top 4.8% in 1942. Therefore, in this period, we
cannot construct the income composition for specified top income groups such as the top 1%. In post-
WWII period, the annual report does not allow us to construct the top income composition before
2007.
Figure 8 presents income composition of total taxpayers who filed composite income tax return in
1933-1942. It also exhibits the income composition by income groups in 2010. Before the WWII,
three major components of income were business income, wage income and rents, the sum of which
accounts for more than 95% of total income during that time. The share of rents declined throughout
1933-1942 while the shares of other two types of income rose. Rents at that time mostly consisted of
farm rents paid to landlords. Even if we take into consideration that some part of interest income
classified as class II income tax was not counted, the share of financial income was insignificant
compared with other three types of income.
Top income composition in 2010 shows that the upper income fraction tends to have the lower
share of wage income. It is noteworthy the share of dividends increases faster for the upper income
fractile in the top income group. It contrasts with the share of interest income which stays constant
regardless of income level.
0
5
10
15
20
25
19
13
19
17
19
21
19
25
19
29
19
33
19
37
19
41
19
45
19
49
19
53
19
57
19
61
19
65
19
69
19
73
19
77
19
81
19
85
19
89
19
93
19
97
20
01
20
05
20
09
US
UK
Korea
Japan
France
11
<Figure 8> Top Income Composition, 1933-1942 and 2010 in Korea
Source: Authors’ calculation.
Notes: 1) The business income in 2010 includes rents.
2) The income composition for 1933-1942 is computed for total taxpayers in the tabulation of class III
income tax who account for the top 1.4% in 1933 to the top 4.8% in 1942.
Comparing income compositions of top income groups in 2010 and the pre-WWII period, we find
the share of farm rents which was large in the pre-WWII period became nil in 2010 and those of wage
income and financial income went up instead. According to the findings from top income
compositions in U.S. and Japan, financial income, particularly dividends which took a large share of
top incomes, fell sharply due to the negative effect of the Great Depression and WWII. It is different
from the case of Korea where farm rents accounted for most capital income and dividends were
insignificant. It reflects the backwardness of Korea in industrial development compared with other
industrialized countries.
3. Understanding the Evolution of Income Concentration in Korea
We have found in the previous section that (i) income concentration in Korea was as high as those
in other countries in the pre-WWII period during 1933-40; (ii) it fell sharply after WWII and remained
low during rapid industrialization until the mid-1990s; (iii) income concentration has increased last 15
years; (iv) as a result, in terms of top income shares, Korea is now higher than Japan and France
although it is lower than the U.K and the U.S. In this section, we explore some plausible explanations
for the evolution of income concentration in Korea.
First, how can we explain a sharp fall of income concentration after WWII? High income
concentration in the pre-WWII period reflects the characteristics of the Korean economy in colonial
period. The Japanese colonizers in Korea played a dominant role with their comparative advantage in
capital and technology. For instance, the Japanese in Korea contributed about 90% of starting capital
to corporations in Korea and thus most executive officers in large corporations were Japanese. The
Japanese in Korea that accounted for 2.9% of population in Korea (in 1940) also owned a sizeable
area of land as large as 9.5% of cultivated land in Korea (in 1942). It implies that many Japanese
landlords in Korea were big landowners. Although we cannot provide statistical evidence from
income tax statistics that does not distinguish ethnicity of taxpayers, it is no doubt that the Japanese
accounted for a large share of top income groups at that time.
After liberation, these Japanese colonizers withdrew from Korea and left their assets which were
vested in the U.S. Military Government in 1945-1948, transferred to the Korean government, and
finally distributed to the private sector during the 1950s. The withdrawal of Japanese entrepreneurs as
well as the severing of the Korean economy from the Japanese economic bloc made a disruptive
impact on the Korean economy. In addition, the division of country into two nations and ensuing
12
political chaos brought economic disorganization in South Korea. Vested properties of Japanese were
managed by officially designated managers until they were transferred to the Korean entrepreneurs
depreciated in both nominal and real value. Therefore, it is not plausible the Koreans who would
replace the Japanese in top income groups after liberalization earned as much as the Japanese.
Income inequality among Koreans at that time may be attributable to inequality in landownership.
During colonial period, the landownership was concentrated in the hands of a small number of
landlords and most farmers were small sharecropping tenants in Korea.14
This landownership was de-
concentrated by land reform. According to the Land Reform Act in 1949, the holdings of Korean
landlords owning more than 3 chongbo (7.5 acres) were not allowed and the excessive holdings were
redistributed to tenants. Although the landlords received securities for the land transfer, the real value
of securities sharply dwindled with high inflation. As a result, the landlords who accounted for a large
proportion of the top income group lost their important income source after the land reform. De-
concentration took place in the late 1940s even before the land reform because landlords expected the
land reform and sold out their land widely.
In sum, the land reform in addition to the withdrawal of Japanese and the collapse of economy
immediately after liberalization had a large effect in reducing income concentration after liberalization.
Income concentration in Korea is thought to stay low throughout the 1950s during which Korea
experienced the War and recovered with help of the U.S. aid. After sharp decline after liberalization, it
stayed relatively low throughout 30 years of rapid growth until the mid-1990s. Since then, it shows an
ascending trend with a steep slope compared with countries such as Japan. It is against Kuznets’
inverse-U shape hypothesis that income inequality tends to deteriorate first and improve later along
the development process. How can we explain this?
First of all, the reversal of Kuznets curve in Korea seems related with structural change in the
Korean economy. During the period when income concentration stayed low, rapid economic growth
accompanying drastic expansion of modern industries decreased the labor surplus of the economy
rapidly. With active job creation in modern industries, benefits of growth spilled over to every corner
of the economy including the bottom income groups. Figure 9 presents annual growth rates of
working age population, non-agricultural employment, and population of farm households in Korea.
During the period of high growth, non-agricultural sector expanded employment at the rates as high as
4-8% per annum as Korea pursued industrialization by promoting labor-intensive export industries. It
induced the labor movement from agriculture to other industries such as manufacturing. The decrease
of farm household population by 4-5% per annum led to the end of labor surplus and the ensuing
wage increase in the rural areas. It pushed the mechanization of farming in Korea and increased labor
productivity in agriculture and farm household income.
During the initial period of industrialization, the impact of expanding non-agricultural sectors is too
small to have a considerable effect on farm household income. This is the reason why we often
observe a polarizing pattern of income growth between traditional sector and modern sector during
industrialization. Farm household population continued to increase until the mid-1960s after initial
spurt of economic growth because high population growth overwhelmed the labor drain from the rural
areas. Until the mid-1970s, Korea’s rural areas were overpopulated with surplus of farmers. Although
the land reform abolished tenancy, overpopulation in the rural areas continued until high growth
started to reduce the rural population as figure 9 shows. As rapid growth in Korea actively created
jobs in non-agricultural sectors and induced labor movement out of agriculture, overall earnings of
workers increased in all sectors of the economy.15
In this respect, benefits of economic growth during
this period trickled down to bottom income groups. Thus, Korea could maintain low income
concentration even though it underwent a drastic structural change.
14
Land under tenant farming accounted for 57.9% of total cultivated land in 1940 (67.6% in case of paddy field)
and tenant farmers accounted for 76.4% of farm households. 15
According to the figure 14 we will discuss later, the average wage income of all income groups including
both the top and the bottom increased at the similar rates during the period of high growth in Korea.
13
<Figure 9> Growth Rates of Working Age Population and Non-agriculture Employment in Korea
Notes: 1) The pre-WWII growth rates are for South Korea only.
2) The growth rates of non-agriculture employment and farm households are computed after
obtaining the moving average over 5 years.
Sources: Statistical Yearbook of the Government of Colonial Korea (Chōsen sōtokufu tōkei nenpō),
Statistical Yearbook of Korea (Hankuk Tonggae Yeongam); Annual Report on Economically Active
Population Survey(Kyungjae Hwaldong Ingu Yeonbo)
However, job creation became stagnant after the mid-1990s as displayed in figure 9, which
weakened the trickle-down effect we saw previously in the period of high growth. The growth rate
of non-agricultural employment fell to around 2% in the 2000s. It corresponds with subsiding GDP
growth rate as Korea enters a mature stage of development. In addition to the stagnant economic
growth, the employment inducement effect of economic growth has been weakened since the 1990s as
industrial structure in Korea shifted toward technology-intensive industries from labor-intensive ones.
This trend accelerated as Korea deepened its economic relationship with China after Korea
established diplomatic relations in 1992. In addition, Korean companies relocated labor-intensive
industries and production process to the sites in low-wage countries, which further reduced demand
for unskilled workers in Korea. It resulted in reduction of manufacturing employment since the job
creation in high-technology industries cannot compensate for the job destruction in low-skill labor-
intensive industries. Korea’s manufacturing sector which actively absorbed the labor during the period
of rapid growth, now releases the labor. To summarize, as job creation effect of economic growth
became weakened, polarizing demands for workers with different skill levels put an end to the trickle-
down effect of economic growth which had a favorable effect in income inequality until the mid-
1990s.
This explanation based on the structural change in Korea may well account for deterioration of
income inequality in Korea but may not fully explain increase of top incomes. Next, we will
investigate tax system, particularly the change in tax rates which might have a direct effect in top
income shares.
According to figure 10, the highest statutory marginal tax rate(MTR) denoted as top MTR ascended
fast during the 1960s and the 1970s and hit a peak of 70%. Then it steadily declined to the half of the
peak level.16
Korea was no exception to the neoliberalism trend of public policy many developed
countries adopted in the early 1980. Among them, Anglo-Saxon countries such as the U.S and the U.K
16
The MTRs in figure 10 are those for the wage income since we use tax rates for the wage income for years
before 1975. After introduction of composite income tax system in 1975, income tax rates became identical
among various types of income.
-6.0
-4.0
-2.0
0.0
2.0
4.0
6.0
8.0
19
25
19
30
19
35
19
40
19
44
19
49
19
55
19
60
19
66
19
70
19
75
19
80
19
85
19
90
19
95
20
00
20
05
20
10
Non-agricultural employment
Working age population
Farm household population
14
cut income tax rates drastically in the 1980s. Top MTR is just the highest rate of marginal tax
corresponding to the top income range specified by tax law. With top MTR alone, we cannot evaluate
true tax burden of top income groups until we take into consideration the number of taxpayers in the
group, who should pay their tax at top MTR. For instance, even though top MTR is high, tax burden
would not be high if few taxpayers earn income high enough to belong to the range affected by top
MTR. Thus, we estimate MRT for the average taxpayer in top 0.1% wage income group to see true
burden of tax for additional income. Figure 10 presents MTR for the average taxpayer in the top 0.1%
wage income group (top 0.1% MTR in the figure) after taking into consideration the weight of
taxpayers who pays at the top MRTs.17
<Figure 10> Top 0.1% Wage Income Shares and Marginal Tax Rate in Korea
Note: Top income shares (%) in the right scale, MTR (%) in the left scale
As we can expect, the top 0.1% MTR is lower than top MTR since some individuals in the group
do not pay tax at top MTR. It was high in the late 1970s at 54% when the top MTR was 70%. It
continued to fall afterwards and went down to 35% in recent years. How could the declining MTR
affect top income shares in Korea? In theory, the tax cut may have a supply side effect by
incentivizing economic activities of top income groups. In addition, it may reduce tax evasion or
increase reported income for tax returns by pass-through of income from other types of income (for
instance, corporate income) to personal income(Matthews, 2011). The CEOs may intensify their
behaviors to seek rents as MTR goes down(Piketty, Saez, and Stantcheva, 2011). In addition, the
lowering MTR may increase the saving capacity of top income groups which helps wealth
accumulation and thus ensuing income increase.18
17
To compute MTR for the average taxpayer in the top 0.1% wage income, we followed the methodology of
Moriguchi and Saez(2008: Appendix C.3). We decompose the top 0.1% into the top 0.01% and the top 0.1-
0.01% and compute MTRs for the threshold income levels of two subgroups. To compute corresponding
MTRs for the threshold income levels, we use income after deduction and tax schedule by income brackets
after assuming the average household has 4 family members (household head with wage income only, non-
working spouse, two dependent children). Then, we estimate MTR for each subgroup by taking a simple
average of MTRs of upper and lower threshold income levels. Then, we obtain the top 0.1% MTR by
averaging two subgroups’ MTRs weighted by average income level of each group. 18
Although the top 0.1% wage income share rose in Korea, other types of income including capital income
increased much faster than wage income for the group. Thus, the positive effect of tax cut on asset
accumulation may have been substantial in Korea. It may not be true of the U.S. where wage income drove
rising income concentration.
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
0
10
20
30
40
50
60
70
80
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
Top MTR
Top 0.1% MTR
Top 0.1% Income Share
Top 0.1% Wage Share
15
Although we cannot clearly distinguish which factor was important in Korea, the falling MTR
seems to have helped income concentration to rise last 15 years. We can support this argument by
comparing Korea with other countries in figure 11. First of all, a close relationship between declining
MTRs and rising top income shares could be clearly observed in the cross-country comparison. Top
income shares rose faster in the countries where MRTs dropped more. Korea belongs to the group
which experienced fast rising top income shares and relatively large tax cut.
<Figure 11> Changes in Top 1% Pre-tax Income Shares and Top Marginal Tax Rates
Source: Piketty, Saez and Stantcheva (2011) except for Korea based on authors’ calculation.
Note: Changes in top income shares and top marginal tax rates are those in % point between 1975-1979 and
2004-2008.
IV. Top Wage Income Shares in Korea, 1963-1984, 1995-2010
This section explores the evolution of top wage income shares in Korea. Unfortunately income tax
statistics in the pre-WWII period does not provide the tabulation of wage income tax by income
brackets and thus we confine our analysis to the period after WWII.
The share of the top 1% group in wage income was already displayed in figure 4. Its overall pattern
is not so much different from that of top income shares. Top 1% wage income share series stayed low
until 1997 but showed a rapidly rising trend afterwards. Wage income concentration in Korea has
been lower than that of total income throughout the period.19
It implies that non-wage income has
been the more important factor for income inequality than wage income.
Figure 12 decomposes the top percentile into three subgroups: the top 0.1%, top 0.5-0.1%, and top
1-0.5%. Although the top 0.1% wage income share has been lower than shares of other subgroups, it
has reduced the gap with other subgroups since 1997. It implies that wage income of the top 1%
increased faster than those of other subgroups. It is noteworthy that, in figure 6 which displayed the
top income shares series, the top 0.1% share surpassed those of other subgroups. In addition, the gap
. 19
One should note that top wage income shares and top income shares are different in terms of their control
total. The former is relative to total employment and the latter is relative to total adult population. Therefore,
top wage income shares tend to be lower than top income shares since the latter counts economically inactive
population in the denominator.
Korea
16
in wage income shares between top 0.5-0.1% and top 1-0.5% was smaller than that in total income
shares. From this, we confirm that other types of income such as business or financial income
contributed more than wage income to the rise of income concentration in Korea. It is different from
the U.S. where fast growing top wage income was a key factor for rising income concentration.
<Figure 12> Decomposition of Top 1% Wage Income Shares in Korea
Source: Appendix.
Note: Top 0.5-01% denotes the wage income share of the top 0.5% excluding the top 0.1%. Top
1-0.5% denotes wage income share of the top 1% excluding the top 0.5 %.
Figure 13 presents top 1% wage income shares series of Korea, Japan and the U.S. to provide a
comparative perspective.20
The top 1% wage income share in the U.S. was high in the pre-WWII
period, fell sharply during the War, and steadily declined until it started to ascend rapidly in the 1970s.
As a result, the top wage share in the U.S ascended to the higher level than it was in the pre-WWII
period. The high wage income concentration in Japan also declined sharply several years ahead of the
U.S. from 1935 to 1944, rose substantially from 1951 to 1961, then again declined gradually, and
remained low relative to the U.S.21
As a result, wage income concentration in the U.S. became much
higher than that in Japan in recent years.
In terms of top wage income shares, Korea followed Japan with a time lag until the mid-1990s. As
discussed in the previous section, Korea experienced a transition of the economic structure as Japan
did, which is reflected as the temporary rise of wage income concentration in both countries in figure
13. The level of wage income concentration in Korea was similar to the level of Japan until Korea
diverges from Japan in the mid-1990s. Korea’s wage income concentration has been rising rapidly last
ten years with a steep slope similar to the U.S. Figure 13 shows a tendency that the Korean economy
approaches to the pattern of the U.S. which shows a rapidly rising trend of wage income concentration.
20
One should note that the top wage shares in figure 13 are constructed with different definitions of control
total for workers across countries. The top wage groups of the U.S. are relative to the total number of tax units
with positive wages and salaries (Piketty and Saez, 2003). The total number of tax units which exclude female
workers with spouse is different from the total number of individual workers used for Korea and Japan.
However, for Japan(Moriguchi and Saez 2008), only regular workers are counted in control total, which is
different from other two countries which count both regular and temporary workers. Therefore, to be exact,
we cannot directly compare the levels of top wage shares across countries. However, this difference in
definition will not make a significant change to the trends we observe in figure 13. 21
Moriguchi and Saez(2008) interpret the rise and fall of the top wage share in Japan around the 1960s as the
phenomenon caused by the transition of Japan from the labor-surplus to the labor-shortage economy with
rapid industrialization.
-
0.50
1.00
1.50
2.00
2.50
3.00
3.50
19
63
19
65
19
67
19
69
19
71
19
73
19
75
19
77
19
79
19
81
19
83
19
85
19
87
19
89
19
91
19
93
19
95
19
97
19
99
20
01
20
03
20
05
20
07
20
09
top 0.5-0.1%
top 1-0.5%
top 0.1%
17
<Figure 13> Top 1% Wage Income Shares
Sources: Appendix 2; Piketty and Saez(2003); Moriguchi and Saez(2008).
How can we explain this pattern of rising wage income concentration in Korea? Moriguchi and
Saez(2008) point out Japan’s highly developed internal labor market as one of important factors for
the diverging trends between the U.S. and Japan. In Japan, the absence of competitive markets for
corporate executives might have prevented the rise of wage inequality. According to Piketty and
Saez(2003: Figure 11), average compensation of the top 100 CEOs in the Forbes list has increased
much faster than average wage since the early 1970s. It increased by more than 25 times in real terms
from US$ 1.5 million to 40 million (in 1999 price) between 1970 and 1999 while the average wage of
full-time workers in the economy computed from National Income slowly increased in between
US$ 30,000 and 40,000. This contrasting divergence of compensation between the top managers and
average workers did not happen in Japan whose corporate culture emphasizes firm-specific human
capital under the internal promotion system.
Since micro-data on top wage groups such as CEO compensation are not available in Korea, we
instead explore the trend of average wage income by income groups.22
Figure 14 displays average
wage income (in 2010 constant price) of the top decile and the rest bottom 90%. The top decile is
decomposed to the top 0.1%, top 0.1-1%, and top 1-10%. First of all, we find the decline of growth
rate of average wage income in total after the economic crisis in 1997. However, average wage
income series of top income groups does not show a distinguishable stagnation of growth unlike the
bottom 90%s. Until 1985, average wage income of all the groups including the bottom 90% increased
at similar rates. In contrast to this, average wage income diverged after 1997 between the top 10% and
the bottom 90%. Particularly, the growth rate of top 0.1% average wage income is distinguished from
those of other income groups.
Although we cannot specify who belongs to the top 0.1% (16,971 in 2010), we guess most
executive officers in large corporations would belong to this group. After the economic crisis in 1997,
many large corporations in Korea reformed their corporate structure and adopted selected elements of
Anglo-Saxon governance system (Chung et al. 2008: 156-189). Compared with the pre-crisis period,
equity financing became a major method of raising capital for large corporations. Thus, the voice of
shareholders became more influential in corporate governance. Large shareholders asked for further
reform to align the management to investor’s interests. The increasing share of foreign shareholders
accelerated this transition toward Anglo-Saxon style shareholder capitalism in Korea. This trend
increased competitive pressures to the top management in Korea. In addition, large corporations that
22
We may obtain information on compensation of executive officers from the report disclosed by corporations
(http://dart.fss.or.kr) starting from 1999 but it is told the information on pay in the report is not so reliable
enough to construct a consistent series of compensation to the top managers.
18
survived the economic crisis increased their scale and scope in business, which increased demand for
able professional managers. In Korea without mature manager market, large corporations, particularly
Chaebols with many affiliated companies, activated the internal labor market for corporate officers.
Chaebol owners delegated the management to professional officers and controlled them through
monitoring and performance-based rewards. Many CEOs in large corporations received exceptional
pay-raises along with stock options. That is, Korea’s Chaebols developed highly competitive internal
labor market for managers of affiliated companies under the umbrella of family owners. In this respect,
Korea seems located in between the U.S. and Japan in terms of the scope of manager market and
incentive system. This kind of performance-based incentive system for the top management was
disseminated to the lower level of management(Chung et al. 2008: 248-259).
<Figure 14> Average Wage Income By Income Groups in Korea
Note: Average wage income (millions of 2000 Korea won) in log scale .
In sum, the rise of wage income concentration in Korea since the mid-1990s is attributable not only
to rapid wage income increase of top income groups but also to the stagnant wages of the bottom 90%.
We could understand this ascending average wage income of top income groups by pointing out the
performance-based reward system of large corporations in Korea. Now, we will investigate the
situation of the bottom 90%. Unfortunately, top income shares series using income tax statistics does
not help us understand the situation in bottom income groups, although it is useful to investigate
income distribution in the top income groups. Therefore, we use the raw data of the Household
Income and Expenditure Survey to know about the bottom 90%.
However, the household survey, due to the problem of under-coverage and under-reporting we
mentioned in the introduction, tends to distort income distribution, particularly of top income earners.
Therefore, we combine the income tax statistics and the household survey to correct this problem. We
extract wage income of each family member from the household survey by using the information on
wage income and employment status of each family member to make the household survey consistent
with income tax statistics.
Figure 15 displays the distribution of workers by income brackets from the two data sources in
2010. For the brackets of lower wage, the number of workers in the household survey is greater than
that from the income statistics which excludes individuals with wage level under exemption point. On
the contrary, for the income brackets over 60 million won, the household survey underestimates the
size of wage earners. The top wage earners over 300 million won reported in income tax statistics are
totally missed in the household survey. While income tax statistics covers most top wage earners close
19
to complete survey, the household survey do not detect a large proportion of the top wage groups.23
Therefore, we use the household survey after substituting the data from income tax statistics for the
missing top income earners in the household survey.
<Figure 15> Distribution of Workers by Income Brackets, 2010
Note: Number of workers in thousands in the horizontal axis and wage brackets in million won in the vertical
axis. Wage brackets in household survey are aligned to be consistent with those of income tax statistics.
<Figure 16> Average Wage Income By Income Groups in Korea (2)
Source: Appendix; authors’ calculation combining income tax statistics and the household
survey
Figure 16 superimposes the average wage income by quintile on the series of average wage
incomes of top wage groups displayed in figure 14. Because the household survey began to cover
single-person households from 2006, we cannot construct average wage income for earlier years
23
It is true of other years as well as 2010 (Kim and Kim, 2013). It seems that the household survey in Korea
suffers from the bias due to high non-response rate which is 20%. Figure 15 also implies that many
responding households may under-report their income level. Income tax statistics could be used as a reference
point to examine the reliability of the household survey.
1
10
100
1,000
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
top 0.1%
top 1-0.1%
top 10-1%
total
bottom 90%
5th quintile
4th quintile
3rd quintile
2nd quintile
1st quintile
20
covering all the employed workers as we defined to construct top wage income shares. Fortunately,
the Household Consumption Survey in 1996 and 2000 covered single-person households and could be
used for our purpose. Thus, we provide the average wage income by quintile only for selected four
years(1996, 2000, 2006, 2010).
Figure 16 allows us to investigate the wage income trend of bottom income groups in detail in
comparison with top income groups. Average wage income levels of the bottom 40%, 1st and 2
nd
quintile, declined since the economic crisis and those of 3rd
and 4th quintile stayed stagnant. It
contrasts with rising average wage income level of the top 20%, 5th quintile. Even within the 5
th
quintile, average wage income levels diverged between the top and the bottom fraction. It clearly
shows the diverging wage in Korea with rising top wages and declining bottom wages. It is different
from the pattern of the earlier years when average wage income levels of all the income groups
increased steadily at the similar rate. It implies that income inequality of Korea has been deteriorating
last 15 years.
Ⅴ. Conclusion
In this paper, we constructed long-term series of top income shares and top wage income shares in
Korea from 1933 to 2010. From this, we have obtained following main findings on the evolution of
income inequality in Korea.
First, income concentration was very high before WWII when Korea was under colonial rule and
dropped drastically after WWII. This pattern is not so much different from what previous studies
found for other industrialized countries such as Japan and the U.S. However, the cause for the change
could be different. Unlike other countries where falling capital income of top income groups played
an important role, Korea’s case is due to the collapse of the colonial system. First of all, the share of
capital income accrued to the top income group, mostly Japanese before WWII, was small relative to
those of other countries. After liberalization, these Japanese colonizers who accounted for a large
share of the top income group in Korea withdrew by leaving their business in Korea. The land reform
after liberalization reduced the rental income of Korean top income groups who earned most of their
income from land lease and personal business. Thus, collapse of the top income share after
liberalization is attributable to the political chaos from regime change in Korea.
Second, in contrast to Kuznets hypothesis, top income shares series in Korea stayed low throughout
the period of rapid industrialization until it started to ascend in the mid-1990s. Income inequality in
Korea has deteriorated with rapidly rising top income shares after the economic crisis. Korea could
maintain low income concentration during the period of rapid growth because economic growth
accompanied active job creation and the steady increase of wage income. This trickle-down effect was
weakened as Korea entered a mature stage of economic development. With advance of globalization
and deepening economic relationship with low-wage countries such as China, major companies in
Korea accelerated the shift of their major business toward technology-intensive sectors by moving
labor-intensive industries and production process to other countries with abundant labor. It reduced
demand for low-skilled workers while it increased demand for high-skilled professionals. As a result,
average wage income of bottom income groups even declined in real terms last 15 years. Although the
diverging demand for different types of workers in labor market may explain worsening income
distribution in overall economy, it is not enough to explain why top incomes increased so rapidly last
15 years.
Thus, third, we pay attention to two institutional factors for the rise of top incomes, the marginal
tax rate(MTR) and the corporate governance structure. Falling MTR since the 1980s seems to help top
income groups to accumulate capital and increase their capital income. In international comparison,
Korea belongs to the country group with big drop in MTRs and rapid increase in top income shares
since the 1980s. In addition, the changes in corporate governance structure after the economic crisis
contributed to big pay-raises of the top management in large corporations. After the economic crisis,
large corporations in Korea relied on direct capital market instead of bank and adopted some elements
of Anglo-Saxon style corporate system. In Korea without mature outside CEO market, large
21
corporations, particularly Chaebols with many affiliated companies, activated the internal labor
market for corporate officers and intensified competition among the top managers of affiliated
companies through performance-based rewards. In this respect, Korea seems to be located in between
the U.S. and Japan in the scope of the CEO market and incentive system.
In this paper, we have studied the long-run evolution of income inequality in Korea by utilizing the
income tax statistics instead of the household survey. However, top income shares do not provide any
information on the situation of the bottom 90 % income groups. Therefore, in this study, we try to
combine income tax statistics and the household survey to investigate income inequality of whole
income groups including the top income groups. We found that income tax statistics could be a useful
reference point to examine the bias of the household survey. For the household survey of Korea, high
non-response rate and under-reporting could be pointed out as culprits of the bias. It is probable that
other countries have similar problems as Korea, which should be examined further to study the
indices of income inequality in comparative perspective.
22
<Reference >
Atkinson, Anthony B.(2005), “Top Incomes in the UK over the 20th Century,” Journal of the Royal
Statistical Society, 168(2), 325-343.
Bank of Korea, ECOS (http://ecos.bok.or.kr)
Chung, Koohyun et al. (2008), Corporate Management in Korea Last 20 Years, SERI. (in Korean)
Japanese Government General of Korea, Statistical Yearbook of the Government of Colonial Korea(Chōsen
sōtokufu tōkei nenpō),
Kim, Nak Nyeon ed.(2012), National Accounts of Korea, 1911-2010, SNU Press.
Kim, Nak Nyeon and Jongil Kim (2013), “Reexamining Income distribution indices of Korea”, Working
paper, Naksungdae Institute of Economic Research. (in Korean)
Korea Tax Administration, Annual Statistical Report on National Tax (Kuksae Tonggae Yeonbo), various
years.
Korea Ministry of Government Legislation (http://www.law.go.kr/main.html)
Korea Statistical Office, Household Income and Expenditure Survey (Gagae Dongjhyang Josa) , various
years.
Korea Statistical Office, Annual Report on Economically Active Population Survey (Gyungjae Hwaldong
Ingu Josa) , various years.
Korea Statistical Office (1996; 2000) Household Consumption Survey (Gagu Sobi Siltae Josa).
Korea Statistical Office, KOSIS (http://kosis.kr/abroad/abroad_01List.jsp)
Korea Statistical Office, Microdata Service System (http://mdss.kostat.go.kr/mdssext)
Maddison, Angus (http://www.ggdc.net/maddison/maddison-project).
Matthews, Stephen(2011), “Trends in Top Incomes and their Tax Policy Implications,” in OECD, Divided
We Stand: Why Inequality Keeps Rising, OECD Publishing.
Moriguchi, Chiaki, and Emmanuel Saez(2008), “The Evolution of Income Concentration in Japan, 1886-
2005: Evidence from Income Tax Statistics,” The Review of Economics and Statistics, 90(4), 713-734.
OECD(2011), Divided We Stand: Why Inequality Keeps Rising, OECD Publishing.
Park, Kijoo and Nak Nyeon Kim (2011), “Consumer Price Index in Pre-Liberalization Period, 1909-1939”,
Economic Analysis 17(1), 131-168. (in Korean)
Piketty, Thomas, and Emmanuel Saez(2001), “Income Inequality in the United States, 1913-1998,” NBER
Working Paper, No. 8467, Appendix, 34-52.
Piketty, Thomas, and Emmanuel Saez(2003), “Income Inequality in the United States, 1913-1998,”
Quarterly Journal of Economics, 118(1), 1-39.
Piketty, Thomas(2003), “Income Inequality in France, 1901-1998,” Journal of Political Economy, 111(5),
1004-1042.
Piketty, Saez and Stantcheva (2011) “Optimal Taxation of Top Labor Incomes: A Tale of Three
Elasticities,” NBER Working Paper 17616. The World Top Incomes Database(http://g-mond.parisschoolofeconomics.eu/topincomes)
23
<Appendix 1> Top Income Shares in Korea, 1933-2010
Notes : (1) The figures are top income shares combining composite income tax and wage income
statistics. Due to data limitations, the figures for 1979-1985 were not adjusted for the
double-counted taxpayers.
(2) P99.99-100 denotes the proportion of top 0.01% income group in total adult population. The
income share of next top 0.04% which could be denoted as P99.95-99.9 can be calculated
by subtracting P99.99-100 from P99.95-100.
P99.99-100 P99.95-100 P99.9-100 P99.5-100 P99-100 P95-100 P90-100
1933 2.95 6.47 8.85 16.68 21.62
1934 2.78 6.12 8.27 15.49 19.96
1935 2.52 5.54 7.51 14.19 18.26
1936 2.44 5.34 7.25 13.75 17.72
1937 2.30 5.21 6.99 12.90 16.48
1938 2.94 5.82 7.61 13.68 17.46
1939 2.84 5.58 7.39 13.19 16.85 29.34
1940 2.79 5.52 7.28 13.13 16.71 29.01
1979 0.86 1.56 2.06 4.79 7.17 18.66 27.03
1980 0.72 1.43 1.99 4.95 7.47 19.78 28.80
1981 0.60 1.30 1.89 4.71 7.14 19.29 28.63
1982 0.53 1.20 1.78 4.73 7.27 20.05 29.66
1983 0.47 1.17 1.78 4.84 7.46 20.49 30.18
1984 0.55 1.23 1.80 4.76 7.35 19.79 29.19
1985 0.64 1.33 1.91 4.75 7.16 19.40 28.84
1995 0.67 1.40 1.96 4.57 6.88 19.18 29.20
1996 0.72 1.50 2.10 4.87 7.27 19.36 32.64
1997 0.73 1.52 2.14 4.96 7.49 19.80 33.37
1998 0.56 1.23 1.76 4.30 6.58 18.44 31.43
1999 0.72 1.49 2.10 4.97 7.55 19.90 33.45
2000 0.85 1.70 2.37 5.42 8.26 21.47 35.77
2001 0.96 1.91 2.60 5.69 8.63 22.30 36.64
2002 1.07 2.09 2.80 6.22 9.25 23.82 38.42
2003 1.11 2.16 2.87 6.32 9.32 24.38 39.14
2004 1.23 2.34 3.14 6.67 9.74 26.40 39.96
2005 1.36 2.57 3.40 6.99 10.07 25.91 38.18
2006 1.43 2.74 3.63 7.53 10.90 28.57 42.43
2007 1.62 3.07 4.03 8.07 11.41 28.86 42.88
2008 1.59 3.05 4.04 8.13 11.51 29.17 43.38
2009 1.57 3.01 4.00 8.12 11.47 28.98 43.13
2010 1.67 3.20 4.24 8.47 11.93 29.64 43.90
24
<Appendix 2> Top Wage Income Shares in Korea, 1963-2010
Notes : (1) The figures are top income shares using wage income statistics.
(2) The notations are the same as Appendix 1.
P99.99-100 P99.95-100 P99.9-100 P99.5-100 P99-100 P95-100 P90-100
1963 0.28 0.73 1.18 3.70 5.86 19.13 30.98
1964 0.14 0.49 0.81 2.73 4.66 14.48 23.80
1965 0.14 0.49 0.85 2.72 4.61 14.72 24.09
1966 0.26 0.63 0.99 2.89 4.62 15.22 24.69
1967 0.27 0.64 1.00 2.92 4.36 12.70 22.62
1968 0.33 0.75 1.16 3.21 5.19 15.94 26.99
1969 0.30 0.68 1.01 3.04 4.93 15.68 26.46
1970 0.24 0.65 1.02 3.11 5.14 16.06 24.33
1971 0.32 0.70 1.09 3.11 5.06 15.52 24.70
1972 0.26 0.59 0.95 2.85 4.65 14.66 23.79
1973 0.21 0.60 0.96 2.84 4.55 13.82 22.05
1974 0.25 0.72 1.14 3.51 5.41 15.31 24.39
1975 0.27 0.79 1.25 4.28 6.34 16.06 24.38
1976 0.22 0.65 1.03 3.02 5.73 15.09 23.12
1977 0.23 0.66 1.05 3.22 5.28 15.32 23.28
1978 0.27 0.78 1.24 3.77 5.99 17.74 27.50
1979 0.43 0.79 1.23 3.59 5.62 16.90 26.80
1980 0.21 0.60 0.94 3.26 5.29 16.21 25.86
1981 0.25 0.69 1.07 3.36 5.32 16.05 25.61
1982 0.23 0.62 1.02 3.38 5.41 16.54 26.60
1983 0.18 0.58 0.97 3.18 5.14 15.68 25.45
1984 0.18 0.51 0.88 3.18 5.08 15.38 24.65
1985 0.37 0.85 1.24 3.35 5.31 15.92 25.35
1995 0.46 0.93 1.27 3.10 4.89 15.33 25.70
1996 0.24 0.64 0.97 2.92 4.84 15.68 25.56
1997 0.18 0.55 0.89 2.87 4.92 17.17 26.52
1998 0.33 0.77 1.12 3.05 4.97 15.38 25.79
1999 0.34 0.87 1.32 3.57 5.74 17.26 27.64
2000 0.37 0.92 1.39 3.69 5.98 18.34 28.24
2001 0.39 0.94 1.39 3.49 5.77 18.01 28.01
2002 0.40 0.97 1.44 3.76 6.08 19.12 29.39
2003 0.36 0.90 1.34 3.61 5.78 18.74 28.77
2004 0.41 1.02 1.45 3.86 6.10 19.06 31.73
2005 0.57 1.25 1.75 4.18 6.49 19.54 32.08
2006 0.61 1.35 1.89 4.48 6.88 21.19 33.88
2007 0.79 1.61 2.19 4.87 7.31 20.92 33.86
2008 0.71 1.54 2.14 4.69 7.30 20.97 33.93
2009 0.69 1.49 2.08 4.62 7.14 20.50 33.19
2010 0.68 1.53 2.16 4.94 7.45 21.47 34.18