ECONOMIC [Asian Economic CRISIS Journal 2001, 1998, ANDVol.INCOME 12 No. 1] 15 3]

INEQUALITY IN KOREA

39

Economic Crisis and Income Inequality in Korea*

Kwang Soo Cheong University of Hawaii at Manoa

Quarterly data for urban worker-households in Korea are used to investigate the effect of the recent economic crisis on income and consumption inequality. There is evidence of sharply increasing income inequality while consumption inequality does not show any apparent trend. Interestingly, the income share of the middle class increased substantially relative to the poor but substantially decreased relative to the rich. There is also a substantial increase in between-group inequality (from income decile groups) which has more than offset the minor reduction in within-group inequality, resulting in increased overall income inequality. It is, therefore, believed that the worker households in Korea are undergoing a distinct process of income stratification parallel with the concentration of income after the economic crisis. Keywords: Income inequality, the Gini coefficient and income distribution in Korea. JEL classification codes: D31, D63, O15.

I.

Introduction

The economic crisis which erupted in Korea at the end of 1997 was truly a watershed event in the country’s recent history. Since then, Korea has experienced dramatic socio-economic changes, and income inequality has quickly emerged as a major policy concern. Frequent questions are whether the burden of the economic crisis is being shared fairly between the rich and the poor, and what will be the immediate and long-run effects on income distribution. Proper answers to these questions will not only provide better understanding of the socio-economic consequences of the crisis but also enable guidelines to be introduced for effectively directing relevant policies. This paper aims to closely investigate the recent trend in income inequality in Korea and thereby both to * I am grateful to Young Back Choi, Jonathan Leightner, David Levine and several other participants in the Annual Convention of the Allied Social Sciences Associations in Boston, January 2000 for their useful comments and suggestions, not to mention the two anonymous referees of the journal. It is a matter of course that all errors are mine. I also gratefully acknowledge the financial support from the Center for Korean Studies at the University of Hawaii.

ASEJ15.1C03

39

11/4/05, 6:52 am

ASIAN ECONOMIC JOURNAL

40

suggest answers to the questions above and to provide a useful road map for interested scholars and policy-makers. The Korean economy, which had been achieving remarkable economic growth and industrialization, nosedived as a result of the foreign currency shortage and weak financial infrastructure at the end of 1997.1 The country’s real gross domestic product (GDP) fell by as much as 5.8% in 1998 after the previous year’s 5% increase.2 Comparing the first quarters of the two years, the decrease in real GDP was less than 4% but the gap had widened to over 7% for the second and third quarters before narrowing to about 5% for the two last quarters. As part of the IMF policy package intended to stabilize the foreign exchange rate, interest rates were raised immediately and dramatically after the onset of the crisis. For example, the overnight inter-bank call rate jumped from 14.54% in November to 25.49% in December and, as a consequence, the number of business bankruptcies soared from 1,469 to 3,197 during the same period. The number had further increased to 9,449 in the first quarter of 1998 before it began a gradual fall to 2,801 in the last quarter of the 1998. Therefore, it was not surprising that the unemployment rate, which had been only about 2% in previous years, increased to 2.6% in 1997 then jumped to a record high of 6.8% in 1998. The impact on the earnings of workers was substantial too. Average annual earnings, which had increased by 11.9% in 1996 and 7.0% in 1997, dropped by 2.5% in 1998 before bouncing back with a 12.0% increase in 1999. The Korean economy recovered from the crisis as swiftly as it had succumbed. Krause (2000) refers to the phenomenon as ‘an exceptional V-shaped recovery.’ Signs of economic recovery were already obvious in the middle of 1998 as private consumption and fixed investment rebounded. From the first quarter of 1999, the real GDP and the average earning of workers also began to show increases from the same quarters of the previous year. The unemployment rate and the number of bankruptcies declined rapidly throughout 1999, following the gradual decrease in interest rates during the second half of 1998. There is no doubt that the Korean government played an important role in the rapid recovery. A series of crisis-management plans and policy reforms were implemented in the financial system, corporate sector, labour market, social safety net and government regulation. To support the extended fiscal involvement of the government, a deficit of 5% of GDP was initially targeted in 1998; however, higher-than-expected revenues and a scaling-back of social welfare expenditures as unemployment fell reduced the deficit to 2.7% in 1999 (OECD, 2000). While the literature on the Korean economy has been rapidly expanding since the onset of the crisis, surprisingly, few studies deal with the distributional 1. There have been a number of studies on the causes and effects of the economic crisis. See, for example, Dutta and Kwack (1998). 2. All statistics are quoted from the issues of the Major Indicators of the Korean Economy published by the Korea Development Institute unless otherwise specified.

ASEJ15.1C03

40

11/4/05, 6:52 am

ECONOMIC CRISIS AND INCOME INEQUALITY IN KOREA

41

aspect of the crisis. Moreover, most of them, such as D. Kim (1999), H. Kim (2000), Lee and Rhee (1998) and Moon, Lee and Yoo (1999), are broadly defined studies on the overall social impact of the crisis and, therefore, discuss the effect on income distribution only briefly and often have no formal analysis of trends in income inequality. Yoo (2000) is an exception that directly examines income distribution after the economic crisis; however, the main focus of the study is on the poverty issues and the analysis of income inequality is limited to annual comparison of income classes. There has been no rigorous study of the impact of the economic crisis on income inequality as yet, and this is the gap that I intend to fill with this paper. Employing new methods of measuring and decomposing inequality and quarterly data on income and consumption expenditure of urban worker-households in Korea, I find that income inequality sharply worsened after the crisis while consumption inequality showed no apparent trend. Interestingly, the income share of the middle class (defined as the fourth to seventh income deciles) substantially decreased both in the absolute percentage and relative to the rich (the top two deciles) but substantially increased relative to the poor (the bottom two deciles). In addition, the results from decomposition analysis on income decile groups suggest that the substantial increase in between-group inequality (from income decile groups) has outweighed the slight decrease in within-group inequality, resulting in an increase of overall income inequality. Therefore, the worker households in Korea seem to be undergoing a distinct process of income stratification parallel with the concentration of income. This paper is organized as follows. Sections II and III discuss, respectively, the data and methodology used in this paper. The empirical findings and their implications are considered in Section IV, while the concluding comments are shared in Section V.

II.

Data

Reliable data are an important prerequisite for a successful income study and, in a sense, an income study is only as reliable as the data used. I use the income and expenditure data from the Urban Household Income and Expenditure Survey (UHIES). The survey is conducted monthly by the National Statistical Office (NSO) in Korea, but only quarterly or yearly data are available publicly. The yearly data include more demographic and economic variables than the quarterly data, but I chose the latter since the subject of this paper is the short-term fluctuations in income and consumption inequality. Although the UHIES covers only urban households, there are no better data available for a nationwide income study in terms of survey frequency and reliability. As a matter of fact, the UHIES data have been used in many previous income studies, such as Ahn (1995), and Whang and Lee (1996).

ASEJ15.1C03

41

11/4/05, 6:52 am

ASIAN ECONOMIC JOURNAL

42

The UHIES covers about five thousand representative households living in the seventy-two cities in Korea, excluding farmers’ and fishermen’s households and single-person households. Presumably, it is the urban households that have been mostly affected by recent socio-economic changes and, therefore, using the UHIES data does not seem problematic given the aim of this study. A serious problem lies in the fact that the UHIES collects income data only from worker households but not from self-employed and employer households. One immediate solution to this problem would be generating income data for the excluded households using the consumption function estimated from the worker households under the assumption that workers and non-worker households have identical consumption propensities, as has been done in several previous studies such as Whang and Lee (1996). However, estimation of the consumption function has its own difficulties and, moreover, there have been no studies empirically verifying such an assumption. For this reason, I focus on the data for worker households in this study, which reduces the monthly sample size to about three thousand households. The data set compiled for this study covers thirteen quarters from the first of 1996 to the first of 1999, inclusive. III. Methodological Issues III.1

Unit of analysis

I depart from previous studies by choosing the distribution of disposable income instead of gross income since the former is believed to better approximate household welfare. I also used the distribution of consumption expenditure, which many researchers prefer to income distribution as a better proxy for household welfare. As a measure of economic welfare, I also compute equivalent household income, which is household disposable income adjusted for household size using an equivalence scale ( ϒ ) defined as ϒ = (A + ρK )γ

(1)

where A and K are, respectively, the numbers of adults and children in the household and ρ and γ are parameters. The parameter ρ is the weight assigned to children, and it implements the idea that children do not require as many resources as adults. Setting ρ equal to zero means totally ignoring the presence of children and setting it at one means children and adults are equally treated. The parameter γ denotes economies of scale within the household, implementing the idea that a household’s resource needs are less than proportional to the household size. Following Williams, Weiner and Sammartino (1998), I set both ρ and γ equal to 0.5.3 3. See, for example, Kakwani and Lambert (1998) as a study using equivalent income with different parameter values for ρ and γ.

ASEJ15.1C03

42

11/4/05, 6:52 am

ECONOMIC CRISIS AND INCOME INEQUALITY IN KOREA

III.2

43

Measure of Inequality

I measure the degree of inequality using a variation of the Gini coefficient. The Gini coefficient satisfies several desirable properties for an inequality measure: the Pigou-Dalton transfer principle, Dalton’s population principle and constant relative inequality aversion, and has been ‘generally accepted to be the best single measure of inequality’ (Thomas, 1994, p. 73). In spite of its popularity, the Gini coefficient has few intuitive meanings. It has been interpreted as the degree of relative deprivation ( Yitzhaki, 1979; Hey and Lambert, 1980; Berrebi and Silber, 1985), the covariance between a household’s income and its income rank ( Lerman and Yitzhaki, 1984; Shalit, 1985), and the weighted average of differences between a household’s importance as a member of a society and its importance as an income-receiving unit ( Milanovic, 1994). Recently, Cheong (1999a) proved that the Gini coefficient is a linear transformation of the centre of gravity of income distribution, that is, the relative income rank of the household on which the income distribution is centred. The exact relationship between the Gini coefficient (G) and the center of gravity (CoG) is given by CoG =

1 (1 + G ) 2

(2)

which shows that CoG ranges from 0.5 and 1 as G ranges from 0 to 1. For example, CoG being 0.67 (or equivalently, G being 0.34) means that the distribution of income is centred on the sixty seventh poorest household of the percentile. Thus, an increase in the Gini coefficient implies that the centre of gravity of the income distribution has moved farther away from the middle-ranked household, so that we can conclude that income inequality has increased. The formula for the centre of gravity of an ordered income distribution (with i denoting income ranks of households) is as simple as the following: N

CoG =

i yi Y

∑N i=1

(3)

where yi denotes the income of the i-th poorest household, N the total population, Y the total income (such that Y = ∑ Ni=1 yi ). Due to its computational simplicity, I computed the centre of gravity first and then derived the Gini coefficient from it. The Gini coefficient and centre of gravity are also computed from the estimated Lorenz curves. Although the Lorenz curve can be directly constructed from income data, parametric estimation of the Lorenz curve remains useful and worthwhile in income studies (Ryu and Slottje, 1998, ch. 4) and many researchers have proposed functional forms for such estimation. In this study, I estimated the Lorenz curve using two functional forms: one proposed by Kakwani (1980) and the other proposed by Rasche, Gaffney, Koo and Obst (1980) (RGKO). These two forms are known to perform well in terms of the goodness of fit and

ASEJ15.1C03

43

11/4/05, 6:52 am

ASIAN ECONOMIC JOURNAL

44

the estimated income shares.4 Their functional forms and corresponding formulae for the Gini coefficient are as follows: Kakwani : p − L( p) = apα(1 − p)β

(4)

where a, α, β > 0, and G = 2a B(α + 1, β + 1), where B is the beta function. 1

RGKO : L( p ) = {1 − (1 − p ) α } β

(5)

where 0 < α, β ≤ 1, and G = 1 − α2 B( α1 , β1 + 1). In estimation, I first constructed percentile income data from the raw data by grouping all households into 100 income groups, so that each income group represents 1% of the sample population. This method not only allows sufficient data points for proper estimation of the Lorenz curve but also generates fixed and equally spaced data points.5 III.3

Decomposition of inequality changes

An additional advantage of using the centre of gravity is that it allows for a useful decomposition of inequality changes into two parts: one due to withingroup inequality and the other due to between-group inequality.6 Suppose that an income distribution is partitioned into income groups (or strata) with equal group sizes, such as income deciles. The centre of gravity can be computed for each group by reranking the households in the group, ignoring the income rank assigned in the total population.7 Denoting the center of gravity within the j-th ( poorest) group by CoGj , the number of income groups by K, the total income within group j by Yj , I obtain the following relationship:8 CoG =

K

∑

j =1

1 Yj CoGj + K Y

K

j Yj 1 − Y K

∑K j =1

(6)

The first term is the weighted sum of CoGj ’s with the weight for each group Y being the product of the group’s population share ( K1 ) and income share ( Yj ), 4. See, for example, Cheong (1999b) for a comparative investigation of existing functional forms. 5. In comparison, the Gini coefficient reported by National Statistical Office is computed on the basis of decile groups. 6. The Gini coefficient can be decomposed too since it is merely a linear transformation of the centre of gravity; however, the direct decomposition of the Gini coefficient is more complicated. See, for example, Lambert and Aronson (1993), Sastry and Kelkar (1994), Silber (1989), Wodon (1999) and Yao (1999), for discussions of possible decomposition of the Gini coefficient using income source, geographical location, family structure and occupation as the grouping variables in addition to income range used in this paper. Due to the limitations in the data used, I did not undertake additional decomposition analysis although inequality decomposition using income source or occupation would be meaningful given the purpose of this paper. 7. For example, the poorest household in each group is assigned the income rank 1 no matter which income group it belongs to. 8. See Appendix for proof.

ASEJ15.1C03

44

11/4/05, 6:52 am

ECONOMIC CRISIS AND INCOME INEQUALITY IN KOREA

45

and it measures the contribution of within-group inequality to overall income inequality. The second term is simply the centre of gravity of the distribution of group income, (Y1, Y2, . . . , YK ); that is, it captures the contribution of the between-group inequality to overall income inequality. In other words, overall inequality is expressed as the sum of within-group inequality and between-group inequality minus a constant that depends upon only the number of income groups. Therefore, as long as the same number of income groups is maintained, one can precisely trace what fraction of inequality changes can be attributed to inequality within groups or inequality between groups. I consider decile income groups in this paper, so K is fixed at 10. IV. Discussion of Findings IV.1

Household characteristics

It does not seem that the household size has been noticeably affected by the economic crisis. Table 1 shows that the number of household members remained almost constant from the third quarter of 1997 until the first quarter of 1999, and I obtained basically the same result from the household size in equivalence scale. However, it is suggested that the age profiles of income classes have been affected by the crisis. Table 2 presents the average ages of household heads in each income decile, and it is shown in the table that the average age of household heads of the poorest decile group increased between the third quarter of 1997 and the third quarter of 1998 while that of each of the other decile groups decreased during the same period. In other words, households with old heads have suffered relatively more severe income reduction. Therefore, although some say that the old people have suffered less or have even gained from the appreciation of their financial assets because of the high interest rates during the early crisis period, the numbers in Table 2 show that this is not true, at least for worker households. IV.2

Income and consumption expenditure

Interestingly enough, income and consumption expenditure have been affected differently by the economic crisis. Tables 3 and 4 present, respectively, the average income and consumption expenditure for income decile groups, and show clearly that more than half of these groups had recovered their pre-crisis (nominal) consumption level (that is, the consumption level in the third quarter of 1997) by the first quarter of 1999 while only the richest group had recovered its pre-crisis (nominal) income level.9 9. According to OECD (2000), increasing private consumption fuelled by pent-up demand led to the economic recovery in 1999.

ASEJ15.1C03

45

11/4/05, 6:52 am

ASEJ15.1C03

Decile

11/4/05, 6:52 am

Bottom Ninth Eighth Seventh Sixth Fifth Fourth Third Second Top

96–1

96–2

96–3

96–4

97–1

97–2

97–3

97– 4

98–1

98 –2

98–3

98–4

99 –1

3.3 3.4 3.5 3.6 3.8 3.8 3.9 3.9 4.0 4.2

3.0 3.4 3.5 3.6 3.7 3.7 3.8 3.9 4.0 4.2

3.1 3.4 3.5 3.6 3.6 3.7 3.8 3.9 4.0 4.1

3.0 3.3 3.5 3.6 3.7 3.7 3.7 4.0 4.0 4.0

3.1 3.3 3.4 3.7 3.7 3.7 3.8 3.9 4.0 4.2

3.0 3.3 3.6 3.7 3.6 3.7 3.8 3.9 3.9 4.2

3.1 3.4 3.5 3.7 3.6 3.7 3.7 3.9 4.0 4.1

3.1 3.3 3.5 3.6 3.6 3.7 3.8 3.9 4.0 4.1

3.2 3.3 3.5 3.6 3.5 3.7 3.8 3.9 3.9 4.1

3.2 3.3 3.5 3.5 3.6 3.7 3.7 3.8 4.0 4.2

3.2 3.3 3.4 3.6 3.6 3.7 3.8 3.9 3.9 4.2

3.1 3.2 3.5 3.5 3.6 3.7 3.8 3.8 4.0 4.1

3.1 3.3 3.5 3.5 3.6 3.6 3.7 3.8 3.9 4.1

ASIAN ECONOMIC JOURNAL

46

Table 1 Average Number of Household Members

46

47

Average Age of Household Head

Decile

96–1

96–2

96–3

96– 4

97–1

97–2

97–3

97– 4

98 –1

98 –2

98 –3

98–4

99–1

Bottom Ninth Eighth Seventh Sixth Fifth Fourth Third Second Top

40.6 39.1 37.8 38.1 39.3 38.8 40.2 40.9 42.4 44.6

40.7 38.8 38.7 38.5 38.4 39.3 39.8 41.7 42.8 44.4

41.1 38.9 37.7 39.0 38.9 38.6 40.2 41.4 43.0 44.9

40.8 38.5 37.4 38.4 39.0 40.1 40.7 41.5 42.3 43.6

40.4 38.5 38.1 37.7 39.3 39.0 40.1 41.5 42.6 44.7

40.5 38.4 38.4 38.8 39.3 39.8 40.7 41.9 42.7 45.2

40.7 38.3 38.8 39.5 39.4 39.9 40.4 42.3 43.2 44.8

42.0 38.3 38.5 39.1 40.0 41.2 40.9 41.8 42.7 45.2

40.9 38.5 38.1 38.1 38.2 39.0 39.7 40.3 41.2 43.4

40.7 38.8 37.8 37.5 38.8 39.9 40.3 40.6 41.1 43.1

42.0 38.2 37.8 37.9 38.1 39.7 39.4 40.2 41.9 44.3

41.1 39.2 38.7 38.8 38.9 40.3 40.6 40.8 41.5 43.4

41.3 38.7 38.5 38.2 38.5 39.2 40.0 41.3 41.5 42.9

ECONOMIC CRISIS AND INCOME INEQUALITY IN KOREA

ASEJ15.1C03

Table 2

11/4/05, 6:52 am

47

ASEJ15.1C03 11/4/05, 6:52 am

Decile

96–1

96–2

96–3

96–4

97–1

97–2

97–3

97– 4

98 –1

98 –2

98 –3

98–4

99–1

Bottom Ninth Eighth Seventh Sixth Fifth Fourth Third Second Top

590 967 1183 1382 1591 1822 2078 2408 2906 4438

617 962 1169 1350 1533 1746 1981 2283 2706 4129

670 1052 1291 1512 1746 1985 2262 2608 3120 4525

630 995 1215 1420 1629 1858 2123 2466 2994 4779

639 1035 1288 1510 1745 1987 2266 2624 3160 4760

659 1052 1291 1504 1698 1913 2167 2495 2971 4366

718 1130 1401 1635 1872 2120 2417 2770 3309 4829

662 1051 1283 1484 1690 1911 2185 2510 2991 4395

546 963 1209 1418 1632 1872 2161 2514 3053 4922

502 878 1090 1283 1479 1694 1926 2242 2693 5085

500 883 1104 1306 1505 1721 1990 2343 2844 4767

540 914 1137 1327 1526 1741 1996 2305 2806 4727

531 907 1142 1353 1568 1798 2074 2417 2934 5449

ASIAN ECONOMIC JOURNAL

48

Table 3 Average Quarterly Income ( in 1000 Won)

48

49

Decile

96–1

96–2

96–3

96–4

97–1

97–2

97–3

97– 4

98 –1

98 –2

98 –3

98–4

99–1

Bottom Ninth Eighth Seventh Sixth Fifth Fourth Third Second Top

771 891 1006 1151 1213 1356 1482 1613 1901 2771

712 874 985 1055 1196 1266 1324 1568 1803 2636

749 931 984 1079 1189 1390 1486 1640 1884 2476

799 916 1061 1155 1282 1432 1469 1718 1852 2805

834 960 1086 1205 1295 1384 1587 1718 2055 2742

739 916 1071 1140 1247 1387 1450 1669 1798 2675

802 973 1082 1245 1270 1454 1545 1697 1948 2764

839 960 1123 1225 1244 1367 1514 1627 1802 2529

819 901 980 1094 1144 1348 1455 1571 1810 2518

703 794 938 1024 1092 1169 1324 1368 1608 2274

701 800 900 987 1100 1169 1318 1458 1671 2316

788 923 1019 1114 1227 1338 1430 1604 1817 2411

782 988 1081 1191 1293 1402 1569 1757 2016 2731

ECONOMIC CRISIS AND INCOME INEQUALITY IN KOREA

ASEJ15.1C03

Table 4 Average Quarterly Consumption ( in 1000 Won)

11/4/05, 6:52 am

49

ASIAN ECONOMIC JOURNAL

50

The tables also show that the middle class, defined as middle-income groups (the fourth to seventh decile groups), reduced their consumption most severely in percentage terms while the percentage reductions in income were negatively correlated with group income levels.10 In addition, it is found that the richest households (the top decile) became about 10.3 times as rich as the poorest households (the bottom decile) in the first quarter of 1999 while they were about 6.7 times as rich before the crisis; whereas the richest spent only about 3.5 times as much as the poorest both before the crisis and in the first quarter of 1999. These findings seem to substantiate what many people in Korea are saying, ‘the economic crisis is over – many people now spend as much as before; but income gaps have widened.’ The tendency for income distribution to become more unequal and more concentrated, as hinted by Table 3, becomes evident in Table 5, which presents the relative income shares of decile income groups. The table shows that, in the third quarter of 1998, the income shares of the top three decile groups were higher than before the crisis but that only the richest group continued to see the same phenomenon through to the first quarter of 1999. Furthermore, the increase in the richest group’s share was no less than 3.5% in the third quarter of 1998 while the second highest increase was only 0.26% in the same period. In the first quarter of 1999, the increase in the richest group’s income share jumped to 6.14%. The rapid increase in income concentration is clear. Table 5 also presents the changes in the absolute and relative income share of the middle class after the crisis. It first shows that the absolute income share of the middle class (MID in the table), which had remained steady at about 36% without showing any obvious trend during the sample quarters before the crisis, gradually decreased after the crisis to a level of 32.78% in the first quarter of 1999 with the exception of a small rebound in the last quarter of 1998. It also shows that the income share of the middle class relative to the richest two decile income groups ( MID/(i1 + i2) in the table) substantially decreased from 0.97 before crisis to 0.77 in the first quarter of 1999. These findings of the shrinking income shares of the middle class are in line with the common observation of ‘the collapsing middle class’ by H. Kim (1999) and Yoo (2000). Interestingly, however, the income share of the middle class relative to the poorest two deciles ( MID/(i9 + i10) in the table), which was even showing a tendency to decrease before the crisis, had increased by more than 10% of the pre-crisis level by the first quarter of 1999. This is not only consistent with the aforementioned finding of income concentration after the crisis, but also indicates how severely the low income groups were affected by the crisis. 10. There is no consensus about how the middle class should be defined. Although it is often suggested that the definition should include socio-economic criteria other than income levels, many previous income studies define the middle class on the baisis of absolute or relative income ranges. Following H. Kim (2000), I consider the fourth to seventh deciles as the middle class.

ASEJ15.1C03

50

11/4/05, 6:52 am

Relative Income Shares (%)

51 11/4/05, 6:52 am

Decile

96–1

96–2

96–3

96–4

97–1

97–2

97–3

97– 4

98 –1

98 –2

98 –3

98–4

99–1

Bottom (i10) Ninth (i9) Eighth (i8) Seventh (i7) Sixth (i6 ) Fifth (i5) Fourth (i4) Third (i3) Second (i2) Top (i1) MID (= i4 ~ i7) MID/(i1 + i2) MID/(i9 + i10)

3.27 5.22 6.19 7.05 8.32 9.41 10.45 12.00 14.95 23.14 35.23 0.92 4.15

3.69 5.51 6.73 7.37 8.67 9.31 10.70 12.06 13.85 22.11 36.05 1.00 3.92

3.52 5.34 6.30 7.45 8.83 9.20 10.62 12.16 14.58 22.01 36.09 0.99 4.07

3.31 5.22 6.29 7.36 8.22 9.15 10.33 11.72 14.68 23.71 35.05 0.91 4.11

3.22 5.11 6.23 7.26 8.23 9.77 10.54 12.50 14.41 22.75 35.79 0.96 4.30

3.47 5.69 6.72 7.46 8.55 9.72 10.63 12.04 14.19 21.55 36.36 1.02 3.97

3.47 5.40 6.39 7.53 8.40 9.12 10.57 12.42 14.69 22.01 35.62 0.97 4.02

3.44 5.57 6.57 7.37 8.26 9.53 10.45 11.96 14.29 22.55 35.62 0.97 3.95

2.70 5.01 5.83 6.87 7.95 9.17 10.53 12.37 15.12 24.44 34.53 0.87 4.48

2.73 4.70 6.09 6.62 8.01 9.02 10.23 11.80 13.50 27.31 33.88 0.83 4.56

2.72 4.72 5.85 6.77 7.96 8.93 10.05 12.54 14.95 25.51 33.71 0.83 4.53

2.98 5.01 5.65 6.79 8.04 8.95 10.36 11.81 14.96 25.46 34.14 0.84 4.27

2.84 4.54 5.56 6.66 7.54 8.51 10.07 11.75 14.39 28.15 32.78 0.77 4.44

ECONOMIC CRISIS AND INCOME INEQUALITY IN KOREA

ASEJ15.1C03

Table 5

51

ASEJ15.1C03

Relative Consumption Shares (%)

11/4/05, 6:52 am

Decile

96 –1

96 –2

96 –3

96–4

97–1

97–2

97–3

97– 4

98 –1

98 –2

98 –3

98 – 4

99–1

Bottom (c10) Ninth (c9) Eighth (c8) Seventh (c7) Sixth (c6) Fifth (c5) Fourth (c4) Third (c3) Second (c2) Top (c1) MID (= c4 ~ c7) MID/(c1 + c2) MID/(c9 + c10)

5.82 6.57 7.18 8.01 8.66 9.55 10.17 10.97 13.35 19.72 36.39 1.10 2.94

5.83 6.85 7.77 7.89 9.27 9.25 9.80 11.35 12.65 19.34 36.21 1.13 2.85

5.89 7.08 7.20 7.97 9.01 9.65 10.47 11.47 13.20 18.06 37.10 1.19 2.86

5.80 6.64 7.60 8.27 8.94 9.74 9.89 11.29 12.56 19.25 36.85 1.16 2.96

5.92 6.68 7.40 8.16 8.60 9.60 10.41 11.54 13.21 18.48 36.77 1.16 2.92

5.53 7.04 7.92 8.03 8.93 10.02 10.11 11.45 12.21 18.77 37.09 1.20 2.95

5.80 6.96 7.39 8.59 8.54 9.37 10.12 11.40 12.96 18.87 36.62 1.15 2.87

6.16 7.20 8.12 8.60 8.59 9.64 10.23 10.95 12.16 18.34 37.06 1.22 2.77

6.03 6.97 7.03 7.89 8.29 9.83 10.55 11.50 13.33 18.60 36.55 1.14 2.81

5.86 6.51 8.03 8.10 9.07 9.54 10.78 11.03 12.36 18.72 37.49 1.21 3.03

5.81 6.53 7.28 7.81 8.89 9.26 10.17 11.91 13.41 18.92 36.13 1.12 2.93

6.05 7.05 7.05 7.94 9.00 9.57 10.34 11.44 13.49 18.08 36.85 1.17 2.81

5.70 6.75 7.18 8.01 8.49 9.06 10.40 11.66 13.50 19.26 35.95 1.10 2.89

ASIAN ECONOMIC JOURNAL

52

Table 6

52

ECONOMIC CRISIS AND INCOME INEQUALITY IN KOREA

53

In contrast, the concentration trend is not obvious with consumption expenditure. The percentage consumption shares of income decile groups are presented in Table 6. In general, the consumption shares of upper-income groups were over their pre-crisis shares both in the third quarter of 1998 and the first quarter of 1999, and the extent of their increases were relatively small. It is, however, notable that the consumption shares of the second and third richest income groups continued to increase in 1999 while their income shares took a turn downward. Another interesting observation made from Tables 5 and 6 is that the distributions of income shares and consumption shares are remarkably different. The tables show that lower-income groups consume more than proportionally to their income shares while the opposite is true for upper-income groups. Therefore, the between-group income inequality is somewhat reduced at the (current) welfare level while it also means that the between-group inequality of accumulated wealth increases rapidly. IV.3

Inequality trend and decomposition

It was not at all surprising to find out that income inequality has sharply increased as a result of the economic crisis, and the results obtained are seamlessly consistent with each other. First, Table 7 presents the centre of gravity, and it is clear from the table that the CoG for the first quarter of 1999 is higher than that for any quarter of 1998, which is in turn higher than that of any previous quarter, no matter whether it is computed directly from income or equivalence income, and whether indirectly from the Lorenz curve estimated using Kakwani’s form or the RGKO form. The table also shows the centre of gravity derived from the gross-income Gini coefficient computed by Moon, Lee and Yoo (1999), and the annual Gini coefficient officially reported by the National Statistical Office. All corresponding Gini coefficients are presented in Table 8 and Figure 1. It is difficult to assess the distributional impact of the economic crisis without knowing what the inequality trend would have been without the crisis. Unfortunately, there is hardly any close study of income trends in Korea for recent years.11 Although Tables 7 and 8 provide the Gini coefficient and the centre of gravity for the four quarters of 1996, those numbers do not reveal any clear tendency and, moreover, Figure 1 suggests that inequality trends may include cyclical factors. Therefore, it seems only fair to say that this study provides no conclusive evidence on the exact extent to which the economic crisis has affected income inequality. Yet the result of decomposition analysis is revealing. 11. The literature provides the estimates of the Gini coefficient up to 1994. Although the debate on the trend in income inequality during 1980s is not resolved yet, there seems to be a general consensus that income inequality had been falling in the early 1990s. See, for example, Ahn (1995) and Whang and Lee (1996).

ASEJ15.1C03

53

11/4/05, 6:52 am

ASEJ15.1C03

consump eqincome income Kakwani RGKO MLY NSO

96–1

96–2

96–3

96–4

97–1

97–2

97–3

97–4

98–1

98 –2

98 –3

98–4

99–1

0.670 0.639 0.647 0.647 0.647

0.661 0.625 0.634 0.633 0.633

0.663 0.630 0.638 0.638 0.638

0.663 0.639 0.647 0.647 0.646

0.657 0.637 0.645 0.645 0.645 0.650

0.654 0.624 0.633 0.633 0.632 0.641

0.659 0.631 0.639 0.639 0.639 0.644

0.646 0.630 0.638 0.638 0.638 0.641 0.642

0.663 0.653 0.659 0.659 0.659 0.661

0.651 0.661 0.668 0.667 0.665 0.664

0.666 0.657 0.665 0.665 0.664 0.662

0.647 0.653 0.662 0.661 0.661

0.665 0.669 0.675 0.675 0.674

0.645

ASIAN ECONOMIC JOURNAL

54

Table 7 Centre of Gravity

0.658

11/4/05, 6:53 am

Notes: consump = consumption expenditure eqincome = equivalent scale income Kakwani = income (using Kakwani’s form for Lorenz curve) RGKO = income (using Rasche, Gaffney, Koo and Obst’s form for Lorenz curve) MLY = gross income (quoted from Moon, Lee and Yoo) NSO = annual gross income (quoted from National Statistical Office)

54

55

consump eqincome income Kakwani RGKO MLY NSO

Gini Coefficient

96–1

96–2

96–3

96–4

97–1

97–2

97–3

97– 4

98 –1

98 –2

98 –3

98–4

99 –1

0.3392 0.2787 0.2942 0.2939 0.2938

0.3223 0.2507 0.2670 0.2669 0.2662

0.3251 0.2593 0.2756 0.2754 0.2756

0.3264 0.2783 0.2932 0.2930 0.2922

0.3136 0.2743 0.2904 0.2902 0.2900 0.3005

0.3085 0.2489 0.2653 0.2651 0.2644 0.2823

0.3175 0.2623 0.2772 0.2770 0.2773 0.2873

0.2918 0.2603 0.2761 0.2759 0.2756 0.2814 0.2830

0.3265 0.3059 0.3188 0.3184 0.3181 0.3222

0.3024 0.3219 0.3352 0.3349 0.3308 0.3283

0.3312 0.3137 0.3300 0.3294 0.3286 0.3238

0.2945 0.3064 0.3232 0.3228 0.3220

0.3300 0.3375 0.3499 0.3495 0.3478

0.2907

0.3157

ECONOMIC CRISIS AND INCOME INEQUALITY IN KOREA

ASEJ15.1C03

Table 8

11/4/05, 6:53 am

55

ASIAN ECONOMIC JOURNAL

56

Figure 1 Inequality Trends 0.36

Gini Coefficient

0.34 0.32 0.30 0.28 0.26 0.24 96–1 96–2 96–3 96–4 97–1 97–2 97–3 97–4 98–1 98–2 98–3 98–4 99–1 Year-Quarter Income (Direct) Income (Kakwani)

Equivalence Income Income (RGKO)

Consumption Income (MLY)

Table 9 presents not only the decomposition of overall inequality into withingroup inequality (CoGK ) and between-group inequality (CoG(K ) ) but also the direction and rate of their changes. First, the table shows that between-group inequality is higher than withingroup inequality in every quarter as in many other decomposition studies in the literature.12 It is also seen that the share of within-group inequality in the overall inequality increased from each quarter of 1996 to the same quarter of 1997. However, this increasing trend was reversed in the middle of 1998, and its share in the first quarter of 1999 was not only lower than for the same quarter of any other year but was also the lowest amount for all quarters except the last quarter of 1996. It is, therefore, suggested that the economic crisis affected within-group and between-group inequalities differently, resulting in an increased importance for between-group inequality. As regards the inequality changes, the table shows that CoG and CoG(K) move in the same direction for all quarters, implying that an increase (decrease) in overall income inequality always accompanies an increase (decrease) in betweengroup income inequality. The table also shows that this is not necessarily true for within-group inequality. For example, the increase in CoG(K ) exceeds the increase in CoG between the first and second quarters of 1998, while there was a decrease in CoGK. In fact, such overshooting of between-group inequality almost forms a pattern after the onset of the crisis as it is found for four quarters 12. It should be noted, however, that decomposition results are not directly comparable to each other since the levels of between-group and within-group inequalities depend not only upon the distribution of income itself but also upon the grouping variables used.

ASEJ15.1C03

56

11/4/05, 6:53 am

96 –2

96 –3

96–4

97–1

97–2

0.6471 0.2485 0.4986 0.4985

0.6335 0.2417 0.4918 0.4914 −0.0136 −0.0069 −0.0067 49.57%

0.6378 0.2449 0.4929 0.4968 0.0043 0.0032 0.0011 25.43%

0.6466 0.2457 0.5009 0.4904 0.0088 0.0008 0.0080 91.31%

0.6452 0.2480 0.4972 0.4989 −0.0014 0.0024 −0.0038 270.33%

0.6327 0.2439 0.4887 0.4991 −0.0126 −0.0041 −0.0084 67.22%

97–3

97– 4

98 –1

98 –2

98 –3

98–4

99 –1

0.6650 0.2539 0.5111 0.4969 −0.0026 0.0014 −0.0040 153.12%

0.6616 0.2546 0.5070 0.5021 −0.0034 0.0006 −0.0040 118.64%

0.6750 0.2552 0.5197 0.4911 0.0134 0.0006 0.0127 95.14%

57

96 –1 CoG CoGK CoG( K) CoGKK ^CoG ^CoGK ^CoG(K) ^^CoG( K)

11/4/05, 6:53 am

Notes: CoG = Overall Inequality CoGK = Within-group Inequality CoG( K) = Between-group Inequality CoGKK = ratio of CoG K to CoG(K) ^CoG = absolute change in CoG ^CoGK = absolute change in CoGK ^CoG(K) = absolute change in CoG(K) ^^CoG( K) = ratio of ^CoG( K) to ^CoG

0.6386 0.6381 0.2471 0.2476 0.4915 0.4905 0.5028 0.5048 0.0060 −0.0006 0.0032 0.0004 0.0028 −0.0010 46.25% 181.07%

0.6594 0.6676 0.2533 0.2526 0.5061 0.5150 0.5006 0.4904 0.0214 0.0082 0.0058 −0.0008 0.0156 0.0090 73.04% 109.35%

ECONOMIC CRISIS AND INCOME INEQUALITY IN KOREA

ASEJ15.1C03

Table 9 Decomposition of Overall Inequality Change

57

ASIAN ECONOMIC JOURNAL

58

out of a total of six quarters. The average contribution of between-group inequality during the six quarters is about 122% of the changes in overall inequality. Therefore, it is believed that the severe deterioration in between-group inequality has more than offset the minor improvement in within-group inequality, resulting in an increase of overall income inequality. In other words, the results strongly suggest that the worker households in Korea are undergoing a distinct process of income stratification parallel with the concentration of income. V. Concluding Remarks Although anecdotal evidence suggests that income distribution in Korea has become more unequal since the onset of the recent economic crisis, there have been few serious studies on the distributional impact of the crisis. Employing high frequency data and new techniques, this study rigorously investigates such an impact and finds that income distribution has not only deteriorated but is also being stratified. This study has two major limitations, which immediately call for future research projects. One is the task of extending income data to include non-worker households. With such data, one can perform decomposition analysis using occupation as a meaningful grouping variable. This research also has to draw on comparative studies on consumption behaviour of worker and non-worker households. The other task is to estimate the baseline trends in income distribution for the 1990s. This can be only done once the first task is completed; but simply extending the data periods in this study seems a reasonable solution in the short run. No one seems to have a clear vision of how the income trends in Korea will be affected eventually. It could be the beginning of another inverted U-curve or a jump onto a new plateau. Despite this uncertainty, income studies have not drawn much attention from previous researchers. One might say this is due to the lack of a widely acceptable, yet manageable, definition of equity or due to the limited availability of reliable data. However, income inequality is an aspect of economic systems as important as economic growth or efficiency, and we should not turn a blind eye to it. After all, we are often more concerned about how the pie is cut than how big it is. Appendix: Proof of Equation (6) Denote the income rankings of the households in the j-th poorest income group by (Gj−1 + 1, Gj −1 + 2, . . . , Gj), where G0 = 0 and GK = N. Also denote the number of households in each group by n, that is, n = NK . Then CoG can be written as CoG =

ASEJ15.1C03

58

i yi   j =1 i=G j −1 +1 N Y   K



Gj

∑  ∑

11/4/05, 6:53 am

ECONOMIC CRISIS AND INCOME INEQUALITY IN KOREA

59

Since Gj

i y ∑ N Yi = i=G j −1 +1 =

G j −G j −1

i + Gj −1 y i+G j −1 N Y

∑ i =1

G j − G j −1

i yG j −1 + i + N Y

∑ i=1

= =

n N n N

G j − G j −1

∑ i=1

G j −G j − 1

Gj − 1 yG j −1 + i N Y

Yj Gj −1 i yG j −1 + i + ∑ Y i=1 n Yj N Yj Gj −1 Yj CoGj + Y N Y

G j − G j −1

∑ i=1

yG j −1 + i Y

we obtain CoG =

K

n Yj CoG j + Y

∑N j =1

K

Gj −1 Y j Y j =1 N

∑

(7)

Since Gj −1 = n( j − 1), the second term of Equation (7) can be written as follows: K

∑

j =1

n( j − 1) Y j N Y

( j − 1) Y j K Y j =1 K

=

∑

=

∑K

K

j =1

j Yj 1 − Y K

K

Yj

∑Y j =1

k

Since

∑Y Yj

= 1, we obtain Equation (6) from Equation (7).

Q.E.D.

j=1

References *Ahn, Kookshin, 1995, Economic development and income distribution in Korea. Economic Development Review, 1, pp. 53–76. Berrebi, Z. M. and Jacques Silber, 1985, Income inequality indices and deprivation: a generalization. Quarterly Journal of Economics, pp. 809–810. Cheong, Kwang Soo, 1999a, A Note on the Interpretation and Derivation of the Gini Coefficient. Department of Economics Working Paper Series No. 99-1, University of Hawaii at Manoa. Cheong, Kwang Soo, 1999b, A Comparison of Alternative Functional Forms for Parametric Estimations of the Lorenz Curve. Department of Economics Working Paper Series No. 99-2, University of Hawaii at Manoa. Dutta, M. and Sung Yeung Kwack, 1998, Symposium on the Korean financial crisis: causes and challenge. Journal of Asian Economics, 9, pp. 609–670. Hey, John and Peter Lambert, 1980, Relative deprivation and the Gini coefficient: comment. Quarterly Journal of Economics, pp. 566 –573.

ASEJ15.1C03

59

11/4/05, 6:53 am

ASIAN ECONOMIC JOURNAL

60

Kakwani, N. C., 1980, On a class of poverty measures. Econometrica, 48, pp. 437–446. Kakwani, Nanak, and Peter J. Lambert, 1998, On measuring inequality in taxation: a new approach. European Journal of Political Economy, 14, pp. 369–380. Kim, Dae Il, 1999, The Impact on Labor Market of the Crisis in Korea. Mimeo, Seoul National University. Kim, Ho-Ki, 2000, Modernity and Social Change in Korea. Nanam Publishing House, Korea. Korea Development Institute, (monthly issues), Major Indicators of the Korean Economy. Korea Development Institute, Seoul. Krause, Lawrence B., 2000, The aftermath of the asian economic crisis for south korea. Journal of the Korean Economy, 1, pp. 1–21. Lambert, Peter J. and J. Richard Aronson, 1993, Inequality decomposition analysis and the Gini coefficient revisited. Economic Journal, 103, pp. 1221–1227. Lerman, Robert, and Shlomo Yitzhaki, 1984, A note on the calculation and interpretation of the Gini index. Economics Letters, 15, pp. 363–368. Lee, Jong-Wha, and Changyong Rhee, 1998, Social Impacts of the Asian Crisis: Policy Challenges and Lessons. Paper prepared for the Korea Development Institute & United Nations Development Programme Conference. Milanovic, Branko, 1994, The Gini-type functions: an alternative derivation. Bulletin of Economic Research, 46, pp. 81–90. Moon, Hyungpyo, Hyehoon Lee and Gyeongjoon Yoo, 1999, Social Impact of the Financial Crisis in Korea. Mimeo, Korea Development Institute. National Statistical Office, 1998, Annual Report on the Family Income and Expenditure Survey 1997. OECD, 2000, Economic Survey of Korea, 2000: Assessment and Recommendations. OECD, Paris. Rasche, R. H., J. Gaffney, A. Y. C. Koo and N. Obst, 1980, Functional forms for estimating the Lorenz curve. Econometrica, 48, pp. 1061–1064. Ryu, H. K. and D. J. Slottje, 1998, Measuring Trends in U.S. Income Inequality. Springer-Verlag, Berlin. Sastry, D. V. S. and Ujwala R. Kelkar, 1994, Note on the decomposition of Gini inequality. Review of Economics and Statistics, pp. 584 –586. Shalit, Haim, 1985, Calculating the Gini index of inequality for individual data. Oxford Bulletin of Economics and Statistics, 47, pp. 185–189. Silber, Jacques, 1989, Factor components, population subgroups and the computation of the Gini index of inequality. Review of Economics and Statistics, 71, pp. 107–115. Thomas, Desmond R., 1994, Estimation of Gini coefficients in selected (OECS) countries. Social and Economic Studies, 43, pp. 71–93. Whang, Seong-Hyeon and Jung-Woo Lee, 1996, The Problems of Distribution and Policy Issues in Korea. Mimeo, Korea Development Institute. Williams, Roberton, David Weiner and Frank Sammartino, 1998, Equivalence Scales, the Income Distribution and Federal Taxes. Congressional Budget Office Technical Paper Series 1992–2. Wodon, Quentin, 1999, Between group inequality and targeted transfers. Review of Income and Wealth, 45, pp. 21–39. Yao, Shujie, 1999, On the decomposition of Gini coefficients by population class and income source: a spreadsheet approach and application. Applied Economics, pp. 1249–1264. Yitzhaki, Shlomo, 1979, Relative deprivation and the Gini coefficient. Quarterly Journal of Economics, pp. 321–324. *Yoo, Gyeongjoon, 2000, The Changes in Distribution and Poverty after IMF and the Relevant Policy Examples from Foreign Countries. The Korea Development Institute Research Summary 2000–1.

*in Korean

ASEJ15.1C03

60

11/4/05, 6:53 am