1. Introduction GRDP measurement does not appear to exhaus- tively explain regional convergence or divergence Regional issues have been one of the subtlest trends, it is useful in identifying the dynamics of and most persistent issues in , as shown by regional disparities. the terms: the 'East-West Divide' and, more recent- ly, ‘the Capital regions versus the Non-Capital 2. Trends in regional divergence(or regions'. And in spite of a great deal of govern- inequality) in Korea ment effort to solve the issue, regional divergence (or inequality) can still be regarded as one of the The overall trend in regional inequality in Korea most pressing problems in Korea. during 1989∼2000 is depicted in Figure 1. It To more closely examine this situation, the fol- plots two indicators, the population-weighted stan- lowing questions have to be asked: How dynamic dard deviation/mean (WSD/Mean) and the popu- and successful were Korea's regional economies in lation-weighted mean absolute deviation (WMAD), the 1990s? Are differences in their economic per- i.e. weighted coefficients of variation, to measure formances widening or shrinking? To answer these the degree of inequality in per capita GRDP questions, this short paper outlines the overall trends in the economic performances of Korea's provinces during the period 1989∼2000, drawing upon variations in their gross % % 25.0 20.0 regional domestic product (GRDP), 18.0 20.0 which is a measure of the value of all WSD/Mean 16.0 the market and collective sector goods 14.0 15.0 12.0 and services produced in an area or WMAD 10.0 produced by the employed residents 10.0 8.0 of an area. Put more succinctly, it 6.0 measures an area's capacity for eco- 5.0 4.0

2.0 mean absolute deviation) WMAD (Weighted WSD/Mean (Weighted standard deviation/Mean) WSD/Mean (Weighted nomic activities or for involving its res- 0 0 idents in the creation or appropria- 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 tion of wealth. Therefore, while the between 15 of the nation's provinces. (Gyeongnam and provinces are treated as a single region because their data just started to be published % % 25.0 separately in 1998.) The results indi- 20.0 18.0 cate that, except for a reversal dur- 20.0 WSD/Mean 16.0 ing the first few years of the 1990's, 14.0 there was an overall clear increase in 15.0 12.0 inequality from 1989 to 2000, with 10.0 10.0 WMAD 8.0 the WSD/Mean reaching 21.7 in 6.0 2000 from 14.1 in 1989, an 5.0 4.0

2.0 mean absolute deviation) WMAD (Weighted increase of about 53.9 percent, and standard deviation/Mean) WSD/Mean (Weighted 0 0 with the WMAD rising by about 87.1 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 percent from 11.6 in 1989 to 21.7 in 2000. Differences in per capita GRDP can be divided into two parts: differ- ences in productivity, which them- selves reflect differences in physical productivity, prices and earnings, and % % differences in the share of the popu- 7.0 6.0 WSD/Mean 6.0 lation in employment (employment 5.0 rate). More formally: 5.0 4.0 4.0 3.0 3.0 WMAD This disaggregation produces a 2.0 2.0 number of insights into the determi- 1.0 1.0 WMAD (Weighted mean absolute deviation) WMAD (Weighted

nants of regional disparities in devel- standard deviation/Mean) WSD/Mean (Weighted 0 0 opment. To examine the role of these 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 two elements in GRDP per head, Figure 2 plots the variation in region- al productivity and employment rates. In terms of span of the 1997∼98 financial crisis, after which productivity, in the first few years of the 1990's, there were falls in inequality once again. there was a very slight drop in productivity differ- Employment rate differentials reached their lowest entials. Except for this very short phase of declin- record in the mid-1990's. ing productivity differentials, the trend was basi- In sum, it seems clear then that productivity dif- cally upwards, especially since the mid-1990's. ferentials accounted for the overall increase in In regards to the employment rates, there were regional divergence (or inequality) over the 1989∼ short phases of rises and dips in the employment 2000 period in Korea. This implies that they can rate differentials. Specifically, there were initial be regarded as an important determinant of the dif- reductions in inequality until the first half of the ferences in the levels of regional development in 1990's, followed by increases in inequality over the Korea. By comparison, there existed a decline in the variations of the employment rate, except for 2000. (It reached an apex of 144.6 percent in the rising trend posted during the 1997∼98 finan- 1998, gradually declining after that.) Looking at cial crisis in Korea. the other eastern provinces of Korea, Gyeongbuk's per capita GRDP increased from 103.5 percent of 3. Differential trends in regional economic the Korean average in 1989 to 124.8 percent in performance 2000, with a relatively sharp rise after the 1997- 8 financial crisis in Korea. In contrast, and How large were the differences in development showed gradually declining trends in GRDP between Korean provinces, i.e. the performance per head over 1989-2000 and never reached the of regional economies, in 1989∼2000? To nation's average, with a range from 79.0 and 88.1 respond to this question, Figure 3 charts esti- percent in 1989 to 59.0 and 72.2 percent in mates of provincial value added (=GRDP) as a 2000, respectively. percentage of the entire nation's average (see also The western part of Korea, including the Table 1). provinces of Jeonnam, Jeonbuk, and Breaking down the data, over the 1989∼2000 Jeju, also exhibited an underperforming pattern. period, Korea's leading province was Gyeongnam The lone exception was Jeonnam, which experi- (including Ulsan whose separate data has only been enced a sharp rise from 86.5 percent in 1989 to officially available since 1998), which saw its 112.3 percent in 2000, with an apex of 126.2 GRDP per head jump from 133.0 percent of the percent in 1998 (declining afterwards). All other Korean average in 1989 to 134.3 percent in provinces, including Jeonbuk, Gwangju and Jeju,

Notes : 1) Numbers in parenthesis indicate the order of rankings. 2) Ulsan, whose GRDP has been released since 1998 due to its administrative separation, is included in Gyeongnam. never even reached the national aver- age of 100 percent during 1989∼ 2000, except for Jeju in 1989 and % 1992. (av.=100) 150.0 In the so-called capital region, 140.0 Busan Daegu Gyeonggi, and Seoul exhib- 130.0 Incheon ited a gradual decline in GRDP per 120.0 Gwangju Daegeon 110.0 head until the first half of the 1990's. Gyeonggi 100.0 Gangwon Chungbuk Since then, the provinces, excepting 90.0 Chungnam Gyeonggi, have continued to see 80.0 Jeonbuk 70.0 Jeonnam decreases, with Seoul and Incheon Gyeongbuk 60.0 Gyeongnam maintaining a level below the Korean Jeju 50.0 average from 1999 and 1992, 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 respectively. By comparison, Gyeonggi experienced a dramatic increase from 1998. In the northeastern region, the southeastern region, Daegu and Busan record- Gangwon, which is adjacent to the capital region, ed the lowest rate and a below-average rate, never reached the Korean average in the 1989∼ respectively, in the 1990's in regards to GRDP per 2000 period. For its part, the central region of head. Also in the southeast, the provinces of Korea that is comprised of the Chungnam, Gyeongbuk and Gyeongnam, which contain major Chunguk and provinces displayed an industrialized areas such as Ulsan, Masan, increasing GRDP per head pattern in the 1990's, , Pohang, and Gumi, maintained per excepting Daejeon, which saw an overall declining capita GRDPs far in excess of the Korean aver- trend during this time. age, while the provinces of Jeonbuk and Gwangju The most striking feature of this chart (Figure 3) fell below. But the most dramatic change occurred is the remarkable growth trend of both the central in Jeonnam; its per capita GRDP rose impressively part of Korea, except for Daejeon, and of until 1998, probably due to industrialized areas Gyeonggi just from 1998. It seems to imply the such as Gwangyang, Yeocheon, and Yeonsu. southward expansion of the capital region in In order to identify the relative roles of the vari- Korea, revealing the divide between the capital and ations in productivity and the employment rate, the non-capital regions. However, it should also be data for each province were plotted with GRDP per noted that the provinces of Gangwon, Daegu, person employed (productivity) on the horizontal Busan, Jeonbuk, Daejeon, and Gwangju continu- axis and the employment rate on the vertical axis ously underperformed the Korean average over the of the graph. Each variable was measured as a per- 1989∼2000 period. (Incheon and Jeju recorded centage of the Korean average (see Figure 4). the largest declines- see the variation in ranking in The four quadrants show different combinations Table 1). In addition, this chart reveals a trend of of the relative productivity and employment rate accelerating decline for six metropolitan regions: growth (see Table 2). In the top right quadrant, Seoul, Busan, Daejeon, Daegu, Gwangju, and both the productivity and employment rate growth Inchoen. are above average (Type I). The top left quadrant Also revealed in the data of the 1989∼2000 includes regions with below average productivity period was the east-west divide. For example, in growth and above average employment rate growth (Type II). In the bottom left quadrant lies ry to the Type II area in 1994, to the Type III areas where both the productivity and employment area in 1995, and to the Type IV area in 1998. rate growth are below average (Type III). Lastly, Overall, it relied upon its productivity lead, except the bottom right quadrant encompasses regions in the mid-1990's, while its employment rate dete- with above average productivity growth and below riorated gradually, ranging from 103.4 percent in average employment rate growth (Type IV). 1989 to 98.0 percent in 2000, with its lowest Most areas of the nation moved across quadrants level coming in 1998. These results indicate that during the 1990's. For instance, the location of the strong positive impact of productivity on rela- Seoul during the 1990's depended upon its pro- tive GRDP per head was in part offset by the ductivity lead in the first half of the 1990's and its downward pressure exerted by the relatively low employment lead in the second half of the 1990's. rates of employment, especially from 1998. Seoul started in the Type IV region, and moved Showing a contrasting trend, the major determi- through the Type I region in 1990 to a Type II nant of Incheon's move from the Type IV to the classification in 1995, which indicates that Seoul Type III category in 1996 was a decline in its level sustained above-average productivity and employ- of productivity to below the national average. For ment rate growth except in the first half of the Gangwon, its relative employment rate growth 1990's. Gyeonggi moved from the Type I catego- exceeded its relative productivity growth over the

Note : Numbers in parenthesis indicate the respective years in which a shift to one of the quadrants took place. 1989∼2000 period, thus shifting it to the Type in 1992 from the Type II region. During this time, II region in 1998 from the Type III region. Jeju stayed entirely in the Type II area, as its In Gyeongbuk, its employment rate growth employment rate was above average while its pro- recorded above the nation's average during the ductivity was below average. This movement may 1989∼2000 period. At the same time, however, partially be the work of differentials in the sectoral its productivity growth fell below the national aver- profiles of different regional economies. age until 1993, motivating its move to a Type I In Chungnam, the key determinant of its sub- classification in 1994 from a Type II one. par GRDP per head was its low productivity in In Gyeongnam, the major determinant of its the early years of the 1990's. But since the mid- nation's-best GRDP per head was its high pro- 1990's, it has been situated in the Type I quad- ductivity. It also saw a relative decline in its employ- rant on account of an employment lead, except ment rate to under the national average from in 1996 and 1997. Chungbuk had a similar pat- 1994, which was behind its move to the Type IV tern with Chungnam as the major determinant of quadrant from the Type I area. In the cases of its below-average per capita GRDP in 1989-1992 Daegu and Busan, the key determinant of their was low productivity. It moved to the Type IV less-than-average per capita GRDP was their low region in 1995 from the Type I region, and then productivity. More specifically, their productivity shifted through the Type I area in 1998 to the plummeted to 62.2 percent and 74.4 percent in Type IV region in 2000. There was a very slight 2000 from 88.7 percent and 93.4 percent in decline in its employment rate from 102.3 per- 1989, respectively, while their employment rates cent in 1989 to 100.1 percent in 2000, while fluctuated with a range of 88.3 to 97.3 percent. there was a remarkable rise in its productivity from Thus, they stayed in the Type III region in 1989- 92.8 percent to 122.7 percent during the same 2000. time. For Daejeon, the main determinants of its In Jeonnam, its employment rate determined its less-than-average GRDP per head in the 1990's location in the Type II area until 1993. However, were low employment rates in the first half of the a strong rise in productivity from 78.9 percent to 1990's and low productivity in the second half of 112.3 percent in 1997 occurred, followed by a the 1990's, which contributed to its relocation decline afterwards. As its relative productivity from the Type IV area to the Type III area in growth had surpassed its relative employment rate 1993. growth in the first few years of the second half of To summarize, the development trajectories of the 1990's, it shifted to the Type I region in 1994 Korea's provinces are shown in Table 3. Type I from the Type II region. For Jeonbuk, the major might be considered a 'growth-sustaining' region, determinant of its subpar per capita GRDP was Type II as a 'structure-upgrading' region, Type III low productivity. That combined with its close-to- as a 'growth-declining' region, and Type IV as a average rates of employment over the 1989-2000 'growth-stagnating' region. The first type implies period led to its remaining in the Type III area. cases where increases in productivity would induce In the case of Gwangju, its level of productivity improvements in employment. The second and declined to 85.0 percent in 2000 from 103.7 per- fourth types refer to cases in which improvements cent in 1989, while its employment rate surpassed in either productivity or employment would be off- the national average over the 1989-2000 period, set by declines in the other. The third type repre- except in 1997 and 2000. This combination of sents situations where there exist declines in both factors led to its relocation to the Type III region productivity and employment. 110.0 120.0

1989 2000 110.0 1989 1989 2000 2000 100.0 100.0 2000 1989 2000 2000 2000 1989 1989 1989 Employment rate (av = 100) 2000 1989 Employment rate (av = 100) 90.0

90.0 80.0 70 80 90 100 110 120 60 80 100 120 140 160 Productivity (av = 100) Productivity (av = 100)

Gyeongbuk Gyeongnam Daegu Busan Seoul Gyeonggi Incheon Gangwon

120.0 120.0

2000 2000 1989 110.0 110.0 1989 2000

1989 1989 100.0 100.0 2000 2000 1989 1989 2000

Employment rate (av = 100) 90.0 Employment rate (av = 100) 90.0

2000 1989 80.0 80.0 70 80 90 100 110 120 70 80 90 100 110 120 130 Productivity (av = 100) Productivity (av = 100)

Jeonnam Jeonbuk Gwangju Jeju Chungnam Chungbuk Daejeon

The nation's provinces can be divided into the region - Non-Capital region' contrasts were found, four types of development as of 2000. Gyeongbuk, with the case of Jeonnam being exceptional. Jeongnam and Chungnam lie in the growth-sus- Second, metropolitan areas seemed to be in eco- taining region. The structure-upgrading region nomic recession, perhaps due to the failure of struc- includes Gyeonggi, Chungbuk and Gyeongnam. tural adjustment in old industries such as textiles Incheon, Gwangju, Daejeon, Daegu, Busan, and and footwear, and the relocation of their facilities Jeonbuk are situated in the growth-declining area. into adjacent provinces or abroad. Also, they Finally, Seoul, Gangwon and Jeju lie in the growth- appeared to fail to upgrade their industrial struc- stagnating region. tures by methods such as attracting knowledge- Some implications can be drawn from this typol- based producer service sectors. Finally, differences ogy. First, there were strong divides in the devel- in productivity have relatively widened development opment of regional economies over the 1989∼ differentials in terms of per capita GRDP, rather 2000 period. Namely, 'East-West' and 'Capital than differentials in employment rates, in Korea. 4. Concluding remarks Incheon and Gwangju displayed poor economic performances probably due to their relocation of Although there is an apparent limit in this short old industries and their failure to upgrade tech- paper to exploring the proximate factors at work nologies and industrial structures to knowledge- reshaping the relative positions of regional based production systems in manufacturing and economies in Korea, some conclusions can be advanced producer service sectors. Finally, there reached. First of all, Korea is characterized by was in fact an expansion of the Capital region strong divergences in the development of its southward, embracing the Chungnam and regional economies, in terms of per capita GRDP, Chungbuk provinces in the 1990's. that is the productive capacity. These divergences This study highlights the overall trends in the are the results of productivity differentials rather divergence between Korea's regions. However, the than differences in employment rates. That is, economic evolution of individual regions was not clearly it is productivity differentials that explain the examined to identify some of its proximate caus- recent overall rise in regional inequality in con- es, which probably include demographic factors temporary Korea, although some of the proximate and changes in productivity in and between dif- causes of this situation have not been identified in ferent industrial sectors, thus limiting this work. detail in this short essay. Second, the metropoli- tan areas including Busan, Taegu, Daejeon,