Evolution of the costs and benefits of business groups: Korean Chaebols between Weak premium, strong discount and strong premium Keun Lee*, Ji Youn Kima, Oonkyu Leeb a School of Economics, Seoul National University, Seoul Korea b Techno-Economics and Policy Program(TEPP), Seoul National University, Seoul, Korea * Corresponding author. Tel.: +82-2-880-6367; kenneth@snu.ac.kr (Keun Lee)

(2) Keun Lee with others, (Econ (2) Keun Lee with others, (Econ. Dev’t & Cultural Change, 2009/3),” Explaining Performance Change of Chaebols over the Two decades: Technological Capabilities vs. Investment Inefficiency” (3) Jeong, M. & Keun Lee, “Sectoral Systems of Innovations and TFP Catching-up by the Korean firms with the Japanese Firms.” > www.keunlee.com

1. Numerous Empirical Research on Business Groups => Inconclusive In earlier studies of the Japanese keiretsu, group affiliation is viewed as beneficial; Hoshi, Kashyap, and Scharfstein (1990, 1991), Prowse (1992) and Ferris, Kumar, and Sarin (1995): keiretsu affiliations lead to reductions in agency, bankruptcy, and monitoring costs as well as liquidity constraints. However, later studies of keiretsu : Weinstein and Yafeh (1998), Morck and Nakamura (1999), and Kang and Stulz (2000); significant costs to group membership due to the presence of an affiliated bank.

2. About Korean Business Groups Author Time Periods Result Other Chang and Choi(1988) 1975~1984 Over performance than stand-alone firm Manufacturing Ind. Choi and Cowing(1999) 1985~1993 Under performance in financial efficiency than stand-alone firm Listed Firm Joh(2003) 1993~1997 Outside auditing F. Lee and Kim(2000) 1996~1999 Under performance in productive efficiency than stand-alone firm Venture Firm Ferris et al.(2003) 1990~1995 Value loss and Under performance than stand-alone firm Group firms performance getting worse in the 1990s. How about post-crisis period? We employ longer-term data to examine the long-term performance of Korean Business Groups in a consistent way

3. Methodology Replication of methods by Ferris et al. (JBF 2003) for 1990~1995 : Chaebol-affiliated firms: lower excess value, profit stability, over-investment, cross-subsidization, larger debt capacity and lower tax burden Diversification discount vs. Value loss We extend to three time periods : 1984~1988 / 1990~1995 / 2001~2003 (including post-crisis; 1998~2000) Firm Excess value = ln [ firm’s actual value / firm’s imputed value ] imputed value = industry median firm value -to-assets ratio (non-chaebol firms) times the firm’s total assets actual value = market value of equity plus book value of debt Chaebol Excess value = ln [ chaebol’s actual value / chaebol’s imputed value] Chaebol’s actual value = Σ (member firm’s actual value) Chaebol’s imputed value = Σ (member firm’s imputed value)

Testing for Several Hypothesis Over-investment hypothesis Cross-subsidization hypothesis Profit stability hypothesis Co-insurance effect (Debt capacity vs. Tax shields)

4. Data Sources Korea Information Service (KIS) Value Plus Korea Securities Research Institute (KSRI) Stock Database Korea Stock Exchange (KSE) Industry Classifications Identification of Chaebol Group Korea Fair Trade Commission (FTC) : Reports of the Top 30 Company Management Efficiency Research Institute : Korea’s Fifty Major Financial Groups Maeil-Business Newspaper : The Annuals of the Korean Firms

Number of Observations Non-chaebol Firm-Year 5. Sample Descriptions Number of Observations Chaebol Firm -Year Non-chaebol Firm-Year Group-Year 1984~1988 295 818 81 1990~1995 759 1,316 173 2001~2003 248 1,369 57

[Table 1] Descriptive statistics: Chaebol firms, non-chaebol firms and chaebol groups Panel A : comparative firm-level characteristics Time Period 1984-1988 1990-1995 2001-2003 Characteristics Chaebol firms Non-chaebol firms Difference Chaebol firms Total assets (billions of won) 379.989 79.287 300.703 *** 946.810 85.820 860.990 2688.659 403.917 2282.742 [221.946] [44.277] 177.669 [458.258] [55.126] 403.132 [1189.772] [135.03] 1054.729 (452.950) (206.653) (1401.866) (87.288) (4929.273) (2591.990) Total debt-to -total asset 0.761 0.716 0.045 0.752 0.633 0.119 0.572 0.549 0.023 [0.777] [0.698] 0.079 [0.754] [0.620] 0.134 [0.573] [0.486] 0.086 (0.118) (0.326) (0.142) (0.267) (0.251) (0.412) Sales 601.155 78.235 522.920 1031.305 75.008 956.297 3365.063 294.817 3070.245 [236.051] [52.032] 184.020 [409.967] [48.730] 361.237 [1103.195] [121.583] 981.612 (1029.413) (145.027) (2031.982) (77.180) (6772.389) (1127.491)

capital expenditure/sales *** Time Period 1984-1988 1990-1995 2001-2003 Characteristics Chaebol firms Non-chaebol firms Difference Non-chaebol irms capital expenditure/sales 0.082 0.038 0.043 0.094 0.052 0.042 *** 0.037 0.001 0.036 [0.038] [0.040] -0.001 [0.051] [0.033] 0.018 [-0.003] [-0.005] 0.002 (0.228) (1.266) (0.321) (0.318) (0.636) (0.482) Current assets /current liabilities 1.166 1.355 -0.189 1.150 1.632 -0.482 1.053 1.870 -0.816 [1.040] [1.236] -0.008 [1.018] [1.492] -0.474 [0.964] [1.425] -0.461 (0.543) (0.552) (0.855) (0.846) (0.600) (1.642) Dividends paid /net incom 0.450 0.439 0.011 0.306 0.309 -0.003 0.223 0.179 [0.362] [0.327] 0.035 ** [0.155] [0.122] 0.033 [0.167] [0.098] 0.070 (1.072) (2.298) (0.583) (1.109) (0.465) (0.381) Market-to- book ratio 1.000 0.989 1.016 1.076 -0.060 0.907 0.840 0.067 [0.974] [0.937] [0.995] [1.012] -0.017 [0.825] [0.757] 0.068 (0.188) (0.331) (0.152) (0.307) (0.340) (0.440) Taxes/sales 0.010 0.020 -0.010 0.008 0.014 -0.006 0.013 0.000 [0.008] [0.015] -0.007 [0.005] [0.009] -0.004 [0.013] 0.005 (0.011) (0.018) (0.012) (0.017) (0.025) (0.171) Beta 0.899 0.576 0.323 1.100 -0.201 0.831 0.606 0.225 [0.887] [0.622] 0.265 [0.878] [1.127] -0.249 [0.870] [0.610] 0.260 (0.909) (1.261) (0.269) (0.304) (0.358) (0.659) Number of observations 295 818 759 1316 248 1369

Group characteristics Panel B : Chaebol-group-level statistics   Time Period 1984-1988 1990-1995 2001-2003 Group characteristics median mean standard deviation Median Number of member firms 2.000 3.148 1.783 4.000 4.483 2.322 3.000 3.983 2.517 Number of industries 2.568 1.183 3.282 1.485 2.474 1.403 Number of firms with negative operating income 0.000 0.086 0.283 0.110 0.332 0.316 0.597 Total assets (billions of won) 594.548 1179.963 1385.453 1974.589 4160.306 4988.491 4257.829 10608.118 14472.653 Total debt-to-total asset 0.779 0.739 0.107 0.777 0.070 0.575 0.563 0.163 Sales (billions of won) 606.110 2006.965 2808.590 1817.438 4528.174 7202.699 3652.717 13763.155 20876.657 capital expenditure/sales 0.033 0.071 0.103 0.065 0.248 -0.018 Current assets/current liabilities 1.130 1.173 0.355 0.977 1.007 0.262 0.886 0.953 0.450 Dividends paid/net income 0.444 0.746 2.155 0.242 0.215 0.403 0.131 0.077 0.463 Market-to-book ratio 0.970 0.973 0.170 0.984 0.990 0.091 0.830 0.887 0.267 taxes/sales 0.007 0.009 0.006 0.008 0.015 0.010 0.020 mean beta 0.883 0.827 0.556 0.920 0.989 0.456 0.825 0.798 0.228 Number of observations 81 173 57 Panel A contains comparative means [medians] (standard deviations) between chaebol and non-chaebol firms. Differences are evaluated using a t-statistics and the Wilcoxon rank-sum test. Statistical significance at the 1%, 5%, and 10% levels are denoted by ***, **, and * respectively. In panel B chaebol group variables are aggregated across member firms and a mean (median) value is estimated. The figures for the 1990-1995 period are taken from Ferris et al (2003).

Measuring the Excess Value vs Measuring the Excess Value vs. Trend of the Excess Value [Table 2] Measuring excess value at the firm and group levels Time Period 1984-1988 1990-1995 2001-2003 Med. Mean s.d. No. Obs. Chaebol Firms 0.007 0.017** 0.144 295 -0.028*** -0.016*** 0.135 759 0.078 0.112*** 0.328 248 Non- chaebol Firms 0.000 0.005 0.186 818 -0.000 0.024*** 0.222 1316 0.019* 0.372 1369 Difference 0.00710 0.0121 -0.0281 -0.041 0.0781 0.0931 Groups 0.018 -0.003 0.129 81 -0.036*** -0.029*** 0.081 173 0.073 0.107*** 0.298 57 Statistical significance level : 1%, 5%, and 10% are ***, **, *, respectively Statistical significant differences level between chaebol and non-chaebol firms : 1%, 5%, and 10% are 1, 5, 10 , respectively.

[Table 3] Periods and Annual regressions of firm excess value Sample Number of Observations (Adj. R2) Intercept Chaebol dummy Leverage EBIT / Sales (Operating income/sales) Capex / sales Beta 1984-1988 1089 -0.311 *** -0.004 0.419 0.107 -0.002 0.006 (0.463) (0.000) (0.718) (0.105) (0.206) (0.141) 1990-1995 2050 -0.189 -0.080 0.350 0.008 -0.044 -0.007 (0.178) (0.889) (0.591) 2001-2003 1617 -0.340 0.066 0.603 0.315 -0.021 0.027 (0.374) (0.002) (0.005) (0.205) (0.173) 1998-2000 1721 -0.176 -0.082 0.389 -0.091 -0.062 -0.060 (0.532) (0.117) (0.007) 1998-2003 3338 -0.243 -0.027 ** 0.433 0.011 -0.034 0.014 (0.423) (0.030) (0.843) (0.265)

(Operating income/sales) Sample Number of Observations (Adj. R2) Intercept Chaebol dummy Leverage EBIT / Sales (Operating income/sales) Capex / sales Beta 1984 187 (0.713) -0.541 (0.000) *** -0.034 (0.013) ** 0.670 0.566 -0.000 (0.873) -0.005 (0.070) * 1985 193 (0.635) -0.550 -0.024 (0.089) 0.705 0.282 (0.032) -0.009 (0.862) 0.023 1986 201 (0.673) -0.384 0.013 (0.526) 0.463 0.130 (0.270) -0.067 (0.170) 0.051 (0.002) 1987 226 (0.494) -0.244 -0.027 (0.159) 0.404 -0.023 (0.754) -0.010 (0.869) (0.012) 1988 282 (0.318) -0.204 0.003 (0.905) 0.314 -0.042 (0.837) -0.046 (0.347) 0.026 (0.007) 1990 324 (0.288) -0.256 -0.008 (0.676) 0.400 -0.051 (0.728) (0.106) -0.007 (0.829) 1991 333 (0.437) -0.319 -0.019 (0.273) 0.446 0.061 (0.663) -0.076 (0.037) -0.060 (0.028) 1992 335 (0.249) -0.227 -0.078 0.328 -0.230 (0.068) -0.039 (0.128) 0.037 (0.192) 1993 338 (0.119) -0.056 (0.312) -0.120 0.221 -0.176 (0.294) -0.124 (0.063) 0.034 (0.359) 1994 386 (0.182) -0.058 (0.183) -0.141 0.266 0.152 (0.230) 0.030 (0.316) 1995 334 (0.127) -0.175 -0.069 0.297 0.498 -0.081 (0.023) -0.064 (0.011)

Number of Observations (Operating income/sales) Sample Number of Observations (Adj. R2) Intercept Chaebol dummy Leverage EBIT / Sales (Operating income/sales) Capex / sales Beta 1998 578 (0.576) -0.238 (0.000) *** -0.092 0.466 -0.044 (0.312) -0.055 (0.002) ** -0.059 (0.049) 1999 577 (0.433) -0.208 -0.062 (0.037) 0.438 0.049 (0.702) -0.020 (0.563) (0.117) 2000 566 (0.601) -0.139 (0.022) -0.085 0.345 -0.231 (0.043) -0.084 (0.012) -0.047 (0.289) 2001 548 (0.534) -0.348 -0.004 (0.881) 0.596 0.253 (0.016) -0.003 (0.903) 0.027 (0.542) 2002 542 (0.332) -0.360 0.031 (0.350) 0.633 0.579 -0.043 (0.306) 0.003 (0.872) 2003 527 (0.314) -0.475 0.141 0.701 0.120 (0.574) -0.005 (0.807) 0.221 This table presents coefficient estimates from regressions for the determinants of firm excess value. The dependent variable is firm excess value which is calculated as the natural log of the ratio of a firm’s actual value (market value of equity plus book value of debt) to its imputed value. Imputed value is the product of the industry median capital-to-assets ratio drawn from a sample non-chaebol firms times the firm’s total assets. The explanatory variables include a chaebol dummy variable (=1 for chaebol firms) and a set of control variables as suggested by Berger and Ofek (1995). Leverage is total liabilities divided by total assets. Firm profitability is calculated as operating income standardized by sales while firm growth is estimated by the capital expenditures to sales ratio. Beta is estimated from the market model using monthly returns. Regression results for each year are also provided. The p-values (parentheses) are reported. Statistical significance at the 1%, 5%, and 10% levels are indicated by ***, **, and * respectively. The figures for the 1990-1995 period are taken from Ferris et al (2003).

Coefficient (p-value) Over-investment Hypothesis and performance Hypothesis Σ(Capital Expenditure/sales of each of its member firms operating in industries whose median Tobin’s q is in the lowest quartile) Higher value of this -> greater investment in unprofitable industries Time Period 1984-1988 1990-1995 2001-2003 Variable Coefficient (p-value) (1) (2) (3) intercept -0.363 (0.081) * -0.462 (0.024) ** -0.475 (0.018) -0.398 (0.000) *** -0.204 (0.398) -0.377 (0.071) -0.379 (0.072) Over-investment 0.079 (0.777) 0.013 (0.961) 0.033 (0.900) -0.347 (0.003) -1.168 (0.086) -1.223 (0.076) -1.286 (0.059) Leverage 0.537 (0.019) 0.582 (0.013) 0.597 (0.010) 0.485 0.463 (0.177) 0.581 (0.085) 0.592 (0.084) Operating income/sales -0.028 (0.933) 0.045 (0.901) 0.039 (0.912) 0.091 (0.651) 2.994 2.806 2.763 Capex/sales -0.015 (0.938) 0.042 (0.821) 0.040 (0.832) -0.066 (0.012) 0.829 (0.083) 0.915 0.974 (0.052) Relatedness -0.092 (0.257) 0.076 (0.014) 0.095 (0.005) 0.029 (0.213) -0.245 -0.082 (0.429) -0.063 (0.673) No. of Obs. (Adj. R2 ) 81 (0.200) (0.206) 173 (0.195) 57 (0.367) (0.313) (0.308)

Cross-subsidization Hypothesis Cross-subsidization measure : Negative Cash-flow (i.e. EBIT < 0 ) ; The effect of chaebol groups’ excess value by a negative cash flow variable Time Period 1984-1988 1990-1995 2001-2003 Variable Chaebol groups (3) Non-chaebol firms Chaebol-groups intercept -0.466 (0.005) *** -0.289 (0.000) -0.321 -0.118 -0.063 (0.807) -0.255 negative cashflow dummy -0.008 (0.846) 0.046 (0.076) * -0.050 (0.027) ** -0.048 (0.086) -0.180 (0.061) -0.133 leverage 0.590 (0.004) 0.408 0.410 0.227 0.427 (0.334) 0.538 Capex/sales 0.062 (0.703) -0.001 (0.438) -0.059 (0.011) -0.041 (0.047) 0.606 (0.201) -0.018 (0.373) Relatedness 0.094 0.013 (0.599) -0.025 (0.889) No. of observations (Adjusted R2 ) 81 (0.217) 818 (0.529) 172 (0.119) 1065 (0.177) 57 1369 (0.360)

Profit stability Hypothesis Time Period 1984-1988 1990-1995 2001-2003 Characteristics Chaebol firms Non-chaebol firm Difference Panel A: Accounting measures of profitability Operating income /total assets -0.013 [-0.013] (0.002) 0.000 [0.000] (0.003) -0.001 *** -0.002 [-0.001] (0.001) 0.002 -0.004 ** 0.015 [0.013] (0.004) -0.012 (0.014) 0.027 0.013 -0.009 Net income/total assets [-0.005] -0.005 -0.003 [-0.003] 0.005 [0.002] -0.008 -0.006 -0.011 [0.003] (0.013) -0.020 (0.234) 0.009 0.003 -0.221 Number of observation 295 818 759 1316 248 1369 Panel B: Monthly stock market measures of return AR(E) mean [median] (variance) [-0.012] (0.018) 0.001 [-0.016] (0.017) 0.004 * [-0.011] (0.010) [-0.008] (0.056) 0.007 [-0.026] (0.254) 0.006 0.021 -0.198 AR(V) [-0.010] (0.019) [-0.014] -0.000 (0.011) 0.017 [-0.002] (0.058) 0.011 [-0.022] (0.256) 0.202 3461 9116 8775 14867 2915 16282 Panel C: Long-run stock market performance: Chaebol firms versus all non-chaebol firms HPR mean 6.939 [6.419] (17.956) 6.010 [4.935] (14.590) 0.929 1.484 3.366 -0.050 [-0.258] (0.452) 0.148 [-0.119] (0.956) -0.139 -0.504 1.424 [0.866] (4.663) 0.581 [0.068] (3.290) 0.843 0.797 1.373 Wealth relative mean 1.132 [1.250] 0.828 [0.842] 1.533 [1.746] 63 126 124 222 86 401

[(1+C.F’s HPR) / (1+N.C.F’s HPR)] Time Period 1984-1988 1990-1995 2001-2003 Characteristics Chaebol firms Non- chaebol firm Difference Panel D: Long-run stock market performance: Chaebol firms versus matched non-chaebol firms HPR mean [median] (variance) 6.939 [6.419] (17.956) 6.206 [5.228] (15.468) 0.733 1.191 2.488 -0.050 [-0.258] (0.457) 0.318 [-0.006] (1.425) -0.368 -0.264 -0.968 *** 1.424 [0.866] (4.663) 1.001 [0.125] (7.056) 0.417 0.740 -2.393 Wealth relative mean 1.102 0.721 1.208 [1.191] [0.738] [1.658] Number of observation 63 61 124 103 86 60 HPR = Wealth relative = [(1+C.F’s HPR) / (1+N.C.F’s HPR)]

Debt-Capacity vs. Tax Advantages through Co-insurance Effect Imperfect Correlation between their cash flows Able to co-insure each other’s debt The debt capacity of chaebol firms should increase ! Increasing the size of the interest tax shields Able to low tax burdens and less tax paid

[Table 6] Chaebols and the Debt-capacity Panel A : Financial leverage summary statistics Time Period 1984-1988 1990-1995 2001-2003 Characteristics Chaebol firms Non-chaebol firms Difference Total debt- to-assets 0.761 [0.777] (0.118) 0.716 [0.698] (0.326) 0.045 0.079 ** *** 0.752 [0.754] (0.142) 0.633 [0.620] (0.267) 0.119 0.134 0.572 [0.573] (0.251) 0.549 [0.486] (0.412) 0.023 0.086 Industry-adjusted leverage 0.038 [0.055] (0.110) 0.015 [0.000] (0.322) 0.024 0.055 0.108 [0.109] (0.127) -0.004 [-0.001] (0.159) 0.112 0.110 [0.013] (0.065) -0.012 0.027 0.013 No. of observations 295 818 759 1316 248 1369 Panel B : Regression result on industry-adjusted leverage No. of Obs. (Adj. R2) Intercept Chaebol dummy Log of total assets Operating income/sales Capex/sales 1984~1988 1113 (0.034) 0.093 (0.726) 0.008 (0.602) -0.000 (0.984) -0.773 (0.000) 0.000 (0.890) 2052 (0.165) -0.319 0.047 0.030 -0.272 0.016 (0.113) 1617 (0.180) -0.011 (0.956) 0.032 (0.215) 0.006 (0.601) -1.193 (0.001) (0.389)

[Table 7] Interest tax shields and taxes paid Panel A : Taxes-paid summary statistics Time Period 1984-1988 1990-1995 2001-2003 Characteristics Chaebol firms Non-chaebol firms Difference Chaebol firms Taxes/sales 0.010 [0.008] (0.011) 0.020 [0.015] (0.018) -0.010 -0.007 *** 0.008 [0.005] (0.012) 0.014 [0.009] (0.017) -0.006 -0.004 0.013 [0.013] (0.025) (0.171) 0.000 0.005 ** Industry-adjusted taxes -0.005 [-0.005] 0.002 [0.000] (0.015) -0.003 [-0.004] 0.003 (0.016) -0.000 (0.024) (0.170) No. of Obs. 295 818 759 1316 248 1369 Panel B : Regression result on industry-adjusted taxes-paid No. of Obs. (Adj. R2) Intercept Chaebol dummy Log of total assets Operating income/sales Capex/sales 1984~1988 1113 (0.141) (0.245) (0.000) -0.001 (0.070) * 0.067 2052 (0.063) 0.016 (0.002) -0.092 (0.838) 1617 (0.114) 0.129 (0.270) 0.009 (0.387) -0.008 (0.289) 0.373 (0.339) -0.070 (0.085)

1984-88 1990-95 2001-2003 +*** - -*** No Yes*** Yes* - / + + Yes Yes** Excess value (firmes : simple mean) [group : regression chaebol dummy ] +*** - -*** Profit stability hypothesis operating income/ total asset net income / total asset short-run equity return (monthly weighted variance) long-run equity return (HPR matched variance) long-run wealth relative (matched) lower variance Indifference higher variance Over-performance lower variance*** Underperformance Over-investment hypothesis No Yes*** Yes* Performance hypothesis - / + + Cross-subsidization hypothesis (NCF) Yes Yes** Debt-capacity advantage (regression result) Tax advantage (regression result)

Korean Business Groups have dramatically changed over the two decades 1984-88 1990-95 2001-2003 - Higher Stock Return, lower accounting profitability with Lower variance Some Chaebol tax Advantage No Overinvestment effect No Performance Impact => Positive Excess value - Lower Stock Return & accounting profitability with Lower Variance Strong Chaebol tax Advantage Strong Overinvestment effect Negative Excess Value - Higher Stock Return & No Chaebol tax Advantage Mild Overinvestment effect Strong Performance Impact => Positive Excess Value

Summary and Concluding Remarks During the post-crisis period, over-investment and diversification hypothesis has no much explanatory power while cross-subsidization has much weakened, and, more importantly, that profitability improvement is the main causes for the value premium associated with group firms. While profit stability hypothesis was true for the 1990s, it was not so after the restructuring as chaebols boast higher profitability with less variation. Chaebols were significantly more levered than non-chaebol firms only during the 1990s, and chaebol firm’s tax shield advantages has now disappeared in 2001-2003, whereas there were some in the pre-crisis period. Implications: Nature of the firms in emerging economies = very dynamic and ever-evolving nature Any conclusive judgment over them based on results from specific short period is misleading

[Appendix] Numbers of Sample Firms included in the regression analysis Time Chaebol Firm (C) Non-chaebol Firm (N) C+N Chaebol Group Nos. of Groups Nos. of Excluded Groups No. of Groups Included in the Analysis 1984-1988 295 818 1113 121 40 81 1998-2000 333 1388 1721 86 14 72 2001-2003 248 1369 1617 78 21 57 1998-2003 581 2757 3338 164 35 129 1984 53 136 189 25 12 13 1985 54 142 196 24 9 15 1986 150 204 23 8 1987 62 170 232 6 18 1988 220 292 5 20 1998 120 458 578 28 1 27 1999 119 577 30 2000 94 472 566 2001 82 466 548 26 7 19 2002 456 542 2003 80 447 527 17

Department of Economics Seoul National University Explaining Performance Change of Chaebols Before and after the Crisis: Technological capabilities vs. Investment Inefficiency Kineung Choo Keun Lee* Keunkwan Ryu and Jungmo Yoon Department of Economics Seoul National University

<Table 3> Production Frontier Functions Estimation for the Three sub-periods Note: Industry dummies are used, but their coefficient estimates are not reported here.

<Table 3> (Continued)

<Table 4> Inefficiency comparison (CMS 1 Chaebols) 1. The t-values are obtained using White’s formula. 2. Positive value of “difference” means that chaebols are less efficient than non-chaebol firms

2 Causes for the Changes: Chaebol vs. non-Chaebol (1) over-investment: use residual from the investment function in the determinants of productive inefficiency equation -> bootstrapping estimation and Hausman-Taylor (2) technological capabilities: patent counts and diversification

<Table5> Chaebol vs. non-Chaebol: over-investment, patents, etc

<Table 7> Determinants of Firms’ Inefficiency (Chaebol Firm definition=CMS1)

<Table 7> (Continued)

Determinants of Productive Efficiency 1 Over-investment tendency was stronger among the Chaebol firms during the first two periods, whereas it became weaker after the 1997 crisis. ->smaller investment inefficiency among the Chaebol firms explains the higher productive efficiency of the Chaebol firms after the crisis.

Determinants of Productive Efficiency 2 “Technological capabilities measured by Patent applications and/or technological diversification,” were not significant for the pre-crisis period but became more significant after the 1997 economic crisis. ->Higher technological capabilities contribute to higher productive efficiency in the post-crisis period.

Summary and Conclusion Korean Chaebols in the 1990s suffered from productive inefficiency arising from inefficient investment drives. Failure of many Chaebols before and during the crisis period implies that only those Chaebols that have succeed in curtailing investment inefficiency and building new technological capabilities have survived the crisis.

Implications 1: Law of decline: right and wrong Results in this paper suggest a need to restate the thesis of institutional or market imperfection in predicting performance of business groups: While market institutions must have affected the performance of all the business group firms, some survived the environmental challenges while others not. No general “law” of long term decline of business groups.

Sectoral Systems of Innovations, Firm-level Learning and TFT Catch-up between the Korean and Japanese Firms Moosup Jung and Keun Lee (School of Economics, Seoul National University) 39

Backgrounds, Motivation, Hypothesis This paper tries to identify the determinants of TFP catch-up by Korean firms with Japanese firms. The degree of catch-up is measured with TFP catch-up index defined in Jung, Lee and Fukao (2008). 40

Definition of TFP catch-up index TFP Catch up index is defined as TFP level difference of each Korean firm from the average level of Japanese firms in the same sector in each year. 41

TFP Catch-up index : All Mannufact. Note : Note TFP level of all Japanese listed firms in each year is set to be 100. We can regard the difference as % gap of TFP between two countries. 42

4 patterns of TFP Catch-up ICPA code industry name 1985 1990 1995 2000 2004 Catch-up pattern 6 Food and kindred products 81.7 110.3 116.7 111.2 110.9 OVER 9 Lumber and wood 124.5 141.1 131.8 137.9 150.9 10 Furniture and fixtures 87.0 99.6 119.2 125.0 129.1 16 Stone clay glass 80.0 92.2 108.9 108.6 112.6 14 Petroleum and coal products 73.7 163.7 195.3 114.0 102.7 JUST 15 Leather 108.5 104.3 128.0 121.1 104.2 18 Fabricated metal 90.7 100.0 128.5 110.0 96.3 19 Machinery non-elect 91.8 92.5 122.0 110.2 20 Electrical machinery 24.0 30.8 75.0 73.1 96.6 22 Transportation equipment and ordnance 74.8 84.0 103.8 97.0 7 Textile mill products 48.8 57.1 81.3 87.8 82.4 UNDER 8 Apparel 7.7 19.4 53.2 57.5 59.6 11 Paper and allied 72.5 75.6 74.0 86.6 21 Motor Vehicles 38.6 54.5 75.1 78.8 88.0 23 Instruments 33.9 40.7 60.2 61.0 12 Printing publishing and allied 81.6 98.4 106.4 111.1 88.3 REVERSE 13 Chemicals 72.7 78.7 91.0 90.0 80.9 17 Primary metal 67.2 70.0 89.2 61.3 24 Rubber and misc plastics 55.6 61.6 80.5 76.0 total 69.5 92.1 86.5 91.2 43

4 catch-up patterns : Shares 44

Just Catch-up: Samsung vs. Matsushita Sam. Elect.: OVER While Industry : JUST 45

Under Catch-up: Hyundai vs. Toyota H.M. : Under like industry 46

learning and capability (numerous literatures) Hypothesis 1, 2 Hypothesis 1 Hypothesis 2 Explicitness of knowledge and technology(+) Degree of embodied technology transfer(+) Top firm dominance(+) Innovation(TFP catch-up) (Sectoral catch-up + Firm level catch-up) Sectoral Systems of Innovation (Malerba 2002, 2004) Firm level learning and capability (numerous literatures) External Discipline (+) Efficiency wage(+) Innovation capability(+) Size(+) 47 47

Explicitness of knowledge and technology (Patent/R&D) and TFP catch-up Industry code (Icpacode) Industry name TFP Catch-up Index in 2004 Explicitness of knowledge and technology (Patent/R&D) 1990 1995 2000 20 Electrical machinery 96.6 6.2 25.6 4.2 21 Motor Vehicles 88 1.4 11.1 2.3 23 Instruments 61 1.2 1.8 2.7 Note. The values in the table are industry total patent number over industry total R&D expenditure(unit : 1billion) in each year. 48

Degree of embodied technology transfer (machinery import ratio) and TFP catch-up Catch-up pattern Degree of embodied technology transfer (machinery import ratio) 1985 1990 1995 2000 2003 Over, Just 0.36 0.32 0.33 0.41 0.42 Under, Reverse 0.34 0.27 0.24 0.25 Note. The values are average of all the sectors(19 mannufacturing sectors:10 sectors with Over and Just patterns and 9 sectors with Under and Reverse patterns) showing each pattern. 49

Regression Result 50

Table 7 Results with the Interaction Terms (market structure and external discipline)

Robustness tests (labor productivity, lagged)  conclusions TFP catch-up of Korean firms is positively related with such sectoral variables as the “explicitness of knowledge and technology” and the “degree of embodied technology transfer”.  helps to explain why the TFP of Korean firms is now close to, or even higher than, Japanese firms in electronics sectors, whereas in automobile sectors, the TFP gaps still remain after some catch-up. Also more catch-up with monopolistic market structure combined with external discipline form world markets.  confirmed extended version of Schumpeterian hypothesis 52