Is Business Group Structure Inefficient? A Long-Term Perspective
Abstract
This paper investigates the long-term value implication of business group affiliation. In order to secure comparability between business-group-affiliated firms and independent firms we employ the matching estimator technique, which selects firms with business characteristics most similar to chaebol firms to create a control group of firms. We find that the long-term performance of chaebol-affiliated firms is superior to that of control firms although the two groups are very similar at the beginning of the sample period. The differential performance between the two groups has been caused by changes in firm characteristics over time. Difference-in-difference estimators on important firm characteristics indicate that, over time, chaebol-affiliated firms become larger and more profitable, grow faster with more investments, have higher debt, and have more foreign ownership, which leads to a larger firm size. Chaebol firms also seem to benefit from tax shield and monitoring effects due to a higher level of debt, and avoid the entrenchment of owner–managers with a higher foreign ownership. However, regressions using difference variables indicate that business group affiliation by itself is not a value-increasing event.
1. Introduction
Business groups are prevalent around the world. They are groups of firms under common ownership and comprise a significant proportion of economic activity in many countries. Claessens et al. (2006) find that, in eight out of the nine Asian countries in their study, the top 15 family groups control more than 20% of the listed corporate assets. In a sample of 13 Western European countries, Faccio and Lang (2002) find that, in ten countries, the percentage of firms controlled by families is greater than 45%.
Korea is not an exception in having business groups in its economy. Korean business groups are called chaebol and their influence is phenomenal. From 1991 to 2001, the size of assets owned by the 200 largest companies in Korea increased from 84.1% to 101.2% of GDP, and sales increased from 70.5% to 86.5%. Of the total assets owned by the 200 largest companies the proportion owned by firms affiliated with the 30 largest chaebol was 67.6% in 1996, 54.3% in 2001, and 69.1% in 2006 (Solidarity for Economic Reform, 2009). Since chaebol influence is huge, the effectiveness of business group structure is vital to the survival of the Korean economy.
Most of the existing papers study governance issues of Korean chaebol. For example, chaebol are often criticized for improper transfer of funds among member companies for the benefit of family members. Bae et al. (2002) investigate this tunneling hypothesis by examining acquisitions made by firms belonging to Korean business groups. They find that while minority shareholders of a chaebol firm making an acquisition lose, the controlling shareholder of that firm benefits because the acquisition improves the value of other firms in the group. Baek et al. (2006) examine whether equity-linked private securities offerings are used as a mechanism for tunneling among chaebol companies and find evidence that chaebol issuers involved in intra-group deals set the offering prices to benefit their controlling shareholders. Joh (2003) suggests evidence of tunneling by showing that resources are often wasted when a business group transfers resources from one member firm to another. Moskalev and Park (2009) argue that tunneling deviates from value-based management.
Tunneling is an example of resource allocation through internal capital markets. Lee et al. (2009) analyze how a financial crisis affects the operation of internal capital markets within a diversified business group. They argue that active internal capital markets allowed business groups to make efficient capital allocations during the early 1990s, but these were barely functional after the Asian financial crisis. Bae et al. (2008) examine “propping,” another example of transaction via internal capital markets, using earnings announcement events made by chaebol firms. They find that the announcement of increased earnings by a chaebol firm has a positive effect on the market value of other non-announcing member firms.
While most existing papers on chaebol analyze corporate governance and intra-group transactions, this paper focuses on the value implication of business groups. Corporate governance and resource allocations through internal capital markets will ultimately have an influence on firm value. Since the Korean economy cannot properly be evaluated without examining chaebol, value creation or destruction by chaebol structure has an important implication of its own. However, there are not many papers on this issue. Although Lee et al. (2009) compare firm value of chaebol-affiliated and independent companies, their empirical results cannot easily be interpreted. Kim and Berger (2009) report that chaebol firms are larger in size, have higher sales growth rates, lower profitability and lower business risk. Since Lee et al. (2009) compare chaebol firms with all other independent firms, their results cannot be interpreted as a chaebol affiliation effect. They may have found a spurious difference in firm value between chaebol and non-chaebol firms, because they compare firms with different business characteristics.
This paper attempts to evaluate the long-term value implication of business groups using the matching estimator technique. To our knowledge there has been no paper that addresses this issue satisfactorily. We will investigate the long-term value implication of chaebol since business group structure is directly related to the success of the Korean economy. In addition, we construct a research design which allows us to perform valid comparisons between chaebol firms and independent firms by selecting the most comparable companies. For this we use Abadie et al.’s (2004) matching estimator technique. This method is relatively new in the corporate finance area. Almeida et al. (2011) apply this technique to study the relationship between corporate debt maturity and the 2007 credit crisis, and Campello et al. (2010) employ the same technique to explain the effects of financial constraints.
In economic inferences, the matching estimator approach has a variety of advantages over ordinary least squares (OLS), since it enables a more accurate comparison between treatment and control groups. First, under OLS, two groups of firms are compared by examining the coefficients of the variable of interest in the regression with control variables included to take care of the difference in firm characteristics. However, if the control variables (covariates) have a poor distributional overlap between the two groups of firms, the controlling becomes very ineffective, leading to a poor regression output. The matching estimator, however, is free of this problem since it selects the closest covariate values when it composes the control group. In implementing the matching estimator technique we select control firms that are most similar to chaebol firms in terms of observable firm characteristics at the beginning of the sample period. We attempt to include a wide variety of firm characteristics that have been argued to have an impact on firm value in the existing literature: firm size, profitability, growth potential, investments, leverage, liquidity, dividend payments, and ownership structure.
Second, OLS is very sensitive to outliers that can produce bias in the estimates of interest. However, the outlier problem is minimized in the case of the matching estimator, since it is basically a non-parametric method. Third, OLS is a linear model, but in some cases non-linear modeling is more appropriate to explain economic phenomena. In other words, the researcher must have a prior understanding of the economic relationship among variables. When using the matching estimator, however, the researcher does not have to have a full-blown model, since the matching estimator merely compares the treatment and control groups and evaluates the difference in outcomes caused by the treatment.
Another advantage of the matching estimator method is the avoidance of selection bias. For example, Chevalier (2000), Campa and Kedia (2002), and Hyland and Diltz (2002) argue that selection bias may explain the diversification discount on the value of business groups, and the evidence of inefficient internal capital markets offered to explain it. Specifically, Campa and Kedia (2002), and Hyland and Diltz (2002), find that the diversified firms trade at a discount to their industry peers prior to becoming diversified, and Chevalier (2000) reveals that the cross-subsidization patterns documented in the literature can be found among merger partners even prior to merging. The matching estimator method naturally minimizes this selection bias problem through the matching process.
Another innovation in our research design is the use of the difference-in-difference estimator. Although we control for observable differences by using matching estimators, there may still be differences between the chaebol group of firms and the control group of firms as the result of unobservable dimensions. In other words, there could be permanent differences between the two groups, as well as bias from comparisons over time in the treatment group that could be the result of trends. The difference-in-difference technique controls differences between the two groups of companies by eliminating differences in unobservable dimensions.
Our results show that there was no difference in firm value between chaebol firms and control firms at the beginning of the sample period, since the chaebol-affiliated and independent control firms were quite similar in firm characteristics. However, chaebol-affiliated firms gained higher firm value as time passed and the difference in firm value grew following the Asian financial crisis. This suggests that business group structure has an advantage over independent firms even though it has many undesirable aspects, too. Since the firm characteristics of chaebol and control firms were quite similar at the beginning of the sample period, the difference in firm value in later periods may have been caused by the changes in underlying firm characteristics. This conjecture is supported by the results of difference-in-difference analyses on firm characteristics. We find that, over time, chaebol firms become larger and more profitable, grow faster with more investments, maintain a higher debt level, and are increasingly owned by foreign investors.
Firm value regressions over the sub-periods indicate that chaebol affiliation has a positive impact on firm value. Firm characteristics that have an impact on firm value include firm size, profitability, investments, research expenses, leverage, dividends, largest shareholding, and foreign ownership. Among these, profitability, investments, research expenses, leverage, largest shareholding, and foreign ownership have a positive impact on firm value, and dividend payments have a negative impact on firm value. Firm size has significantly negative coefficients over the pre-crisis and crisis periods, but has a significantly positive coefficient over the post-crisis period. When we perform a regression analysis with difference variables, the results are consistent with the previous analysis. But, the coefficient on the chaebol dummy is not significant at all over the crisis and post-crisis periods. Since value changes are explained by changes in firm characteristics, not by chaebol affiliation itself, the implication is that chaebol affiliation per se is not a value-increasing event.
This paper is organized as follows. Section 2 explains the data screening, variable definitions, and the matching process using the matching estimator technique. Section 3 shows the effectiveness of the matching procedure by mean tests and Kolmogorov–Smirnov tests for distributional differences. Section 4 presents evidence of the long-term valuation advantage of business group affiliation. Section 5 shows differential changes in firm characteristics between chaebol and control firms over time that may cause differences in firm value. Section 6 reports panel regression results on firm value followed by the analysis of contributing factors on firm value increase in Section 7. Section 8 presents panel regression results using difference variables to figure out the relationship between the change in firm value and changes in covariates. Section 9 concludes the paper with discussions on future research.
2. Data
The financial and ownership data for this paper were collected from the TS2000 provided by the Korea Listed Companies Association. This is a representative database that provides time series accounting data and stock prices at the end of fiscal years. We attempted to collect data with the longest time series since our focus is the value implication of chaebol affiliation in the long term. Although TS2000 provides data from 1981, our sample period runs from 1990 to 2010 since there are many missing observations before 1990. As for the chaebol affiliation dummy, we utilize the classification provided by the Korean Fair Trade Commission (KFTC). The KFTC classifies a business group as a group of companies in which the controlling shareholder and the affiliated companies own more than 30% of the shares. In a business group, the owner–managers have full control over member firms by pyramiding and cross shareholding.1 As a result, the business ties of member firms are so strong that the entire chaebol operates according to the intention of the owner–manager. Firms that are members of these business groups are coded 1 and 0 otherwise.
Initially, 25 300 firm-years were collected, including both Korea Composite Stock Price Index (KOSPI) and Korean Securities Dealers Automated Quotations (KOSDAQ) market firms. From this we deleted observations with unreasonable values for the variables used in our analysis. For example, we deleted observations with foreign ownership greater than 100% (5 880 observations), a debt ratio bigger than one (304 observations), a dividend payout ratio greater than one (461 observations), a dividend payout ratio less than zero (131 observations), and Tobin’s Q bigger than ten (1671 observations); 16 853 firm-years remained after screening. We further deleted observations if any of the variables used in our analysis had missing values.
In order to compare chaebol with non-chaebol firms, we selected the appropriate benchmark firms that were comparable to chaebol firms. For this we employed the Abadie and Imbens (2006) matching estimator technique based on data from the beginning of our sample period (1990). The matching estimator technique selects the most comparable firms by computing differences in covariates between chaebol and non-chaebol firms. As conditioning covariates, we used asset size, leverage, liquidity, asset growth, profitability, shareholdings of the largest shareholder, and shareholdings of foreign shareholders, which are often cited as important determinants of firm value in the existing literature (Hoshi et al., 1991; Gopalan et al., 2007; Chen and Chuang, 2009; Campello et al., 2010). The variables are defined as follows: size = ln(total assets); debt = total debt/total assets; liquidity = current assets/current liabilities; growth = (total assetst – total assetst−1)/total assetst−1; profitability = (EBIT + depreciation)/total assets; big = equity holding of the largest shareholder; and foreign = equity holding of foreign shareholders. In addition, we used a six-digit industry code classified by the Korea Exchange as an exact matching variable. As a measure of firm value we used Tobin’s Q, defined as total assets minus equity plus stock price multiplied by the number of shares outstanding divided by total assets.
For each chaebol firm Abadie et al.’s (2004) matching estimator technique considered all independent firms in the same industry code. If there were no chaebol firms in a particular industry, all companies in the same industry were ignored. If there were chaebol firms in a certain industry, the matching procedure calculated the N-dimensional distance from the chaebol firm based on the covariates of each non-chaebol firm and selected a firm with the shortest distance as a matching firm.
3. Selecting Proper Comparison Group
Since this paper aims to compare the long-term performance of business-group-affiliated firms and independent firms, comparability between the two groups of firms must be secured. In this regard, our paper distinguishes itself from other papers on business groups with the use of an empirical methodology that takes the effect of observable differences into account. The control group of non-chaebol firms is chosen via a matching procedure, to ensure that there are virtually no observable differences between these firms and those in the treated sample of chaebol firms.
The effectiveness of this approach is presented in Table 1. Panel A shows the mean differences of variables between business-group-affiliated firms and independent firms without considering comparability between the two groups. Except for variables representing company growth and shareholdings of foreign investors, variables are statistically significantly different between chaebol and independent firms. As expected, chaebol firms are much bigger than independent firms. They have higher levels of debt and Tobin’s Q values than independent firms. But chaebol firms have lower profitability and liquidity and a lower share ownership by the largest shareholders. If we compare long-term performances of group-affiliated and independent firms, we cannot determine whether the result is caused by business group affiliation since we compare firms with totally different business characteristics. For example, if we compare chaebol member firms with small independent firms, we observe a difference in performances between the two groups of firms not because of group affiliation, but because of firm size. In order to ascertain the effects of group affiliation on long-term performance, therefore, we have to compare firms with similar characteristics except for the chaebol affiliation.
| Q | Size | Debt | Liquidity | Growth | Profitability | Big | Foreign | |
|---|---|---|---|---|---|---|---|---|
| Panel A: Mean difference between treated (chaebol) and non-treated (independent) firms | ||||||||
| Chaebol Firms | 1.0738 | 19.7029 | 0.7230 | 1.1660 | 0.1995 | 0.0681 | 0.2540 | 0.0272 |
| Independent Firms | 1.0098 | 17.9448 | 0.6140 | 1.5671 | 0.1946 | 0.0845 | 0.3116 | 0.0149 |
| Difference | 0.0640 | 1.7581 | 0.1090 | −0.4012 | 0.0049 | −0.0164 | −0.0577 | 0.0123 |
| t-value | 2.0388** | 13.4950*** | 4.9577*** | −4.3607*** | 0.2512 | −2.1401** | −2.9442*** | 1.6211 |
| Panel B: Mean difference between treated (chaebol) and control firms | ||||||||
| Chaebol Firms | 1.0738 | 19.7029 | 0.7230 | 1.1660 | 0.1995 | 0.0681 | 0.2540 | 0.0272 |
| Control Firms | 1.0302 | 19.0543 | 0.7088 | 1.2405 | 0.1912 | 0.0717 | 0.2484 | 0.0226 |
| Difference | 0.0436 | 0.6485 | 0.0142 | −0.0746 | 0.0083 | −0.0036 | 0.0056 | 0.0047 |
| t-value | 1.2584 | 6.5329*** | 0.8439 | −1.8839* | 0.5806 | −0.9028 | 0.5811 | 1.2931 |
Panel B shows the comparison between chaebol and control firms that are most similar to chaebol firms in terms of covariate values. Control firms are selected by employing the matching estimator technique wherein we take into account firm characteristics related to firm valuation that are often cited as important in the existing literature and practice: firm size, short-term solvency, long-term solvency, profitability, growth rate, and corporate governance or ownership structure. We select firms that are closest to chaebol firms in terms of the above covariate values as at 1990, the beginning of our sample period. Except for size and liquidity variables there are no significant differences in firm characteristics between chaebol and control firms. Even for size and liquidity variables the differences between chaebol and control firms in Panel B are much smaller than those between chaebol and independent firms in Panel A. However, there is no difference in firm value as Tobin’s Q values are not statistically different between the two groups. Thus, the matching procedure seems to be very effective in forming a comparable group of firms. Since we start with firms that have very similar characteristics at the beginning of our sample period, any difference in long-term performance can be ascribed to business group affiliation.
Table 2 compares the distributional properties of covariates among treated (chaebol), non-treated (independent), and control firms. It shows the 25, 50, and 75th quintile values and the results of the Kolmogorov–Smirnov test for distributional similarities among treated, non-treated, and control firms for each covariate. As shown in Panel A, distributions of covariates are completely different between chaebol and independent firms. Except for the growth variable, all variables are statistically significantly different. Therefore, if we compare the long-term performance of chaebol firms with that of independent firms, we cannot determine the main driving forces that generate differences in performance since we perform the comparison based on totally different groups of firms.
| 25th % | Median | 75th % | p-value from Kolmogorov–Smirnov test | ||
|---|---|---|---|---|---|
| Panel A: Difference between treated (chaebol) and non-treated (independent) firms | |||||
| Q | Treated | 0.963 | 1.077 | 1.149 | 0.012** |
| Non-treated | 0.881 | 1.027 | 1.159 | ||
| Size | Treated | 19.028 | 19.651 | 20.304 | 0.000*** |
| Non-treated | 17.235 | 17.869 | 18.527 | ||
| Debt | Treated | 0.644 | 0.731 | 0.806 | 0.000*** |
| Non-treated | 0.507 | 0.635 | 0.715 | ||
| Liquidity | Treated | 0.800 | 0.968 | 1.401 | 0.000*** |
| Non-treated | 1.129 | 1.418 | 1.818 | ||
| Growth | Treated | 0.111 | 0.182 | 0.282 | 0.484 |
| Non-treated | 0.088 | 0.169 | 0.282 | ||
| Profitability | Treated | 0.041 | 0.060 | 0.088 | 0.019** |
| Non-treated | 0.051 | 0.077 | 0.120 | ||
| Big | Treated | 0.146 | 0.250 | 0.319 | 0.006*** |
| Non-treated | 0.208 | 0.304 | 0.383 | ||
| Foreign | Treated | 0.000 | 0.000 | 0.016 | 0.001*** |
| Non-treated | 0.000 | 0.000 | 0.000 | ||
| Panel B: Difference between treated (chaebol) and control firms | |||||
| Q | Treated | 0.963 | 1.077 | 1.149 | 0.036** |
| Control | 0.952 | 1.068 | 1.105 | ||
| Size | Treated | 19.028 | 19.651 | 20.304 | 0.000*** |
| Control | 18.519 | 19.005 | 19.547 | ||
| Debt | Treated | 0.644 | 0.731 | 0.806 | 0.178 |
| Control | 0.644 | 0.704 | 0.790 | ||
| Liquidity | Treated | 0.800 | 0.968 | 1.401 | 0.000*** |
| Control | 1.002 | 1.144 | 1.337 | ||
| Growth | Treated | 0.111 | 0.182 | 0.282 | 0.385 |
| Control | 0.091 | 0.193 | 0.250 | ||
| Profitability | Treated | 0.041 | 0.060 | 0.088 | 0.464 |
| Control | 0.046 | 0.069 | 0.088 | ||
| Big | Treated | 0.146 | 0.250 | 0.319 | 0.385 |
| Control | 0.168 | 0.218 | 0.308 | ||
| Foreign | Treated | 0.000 | 0.000 | 0.016 | 0.241 |
| Control | 0.000 | 0.000 | 0.006 | ||
Panel B reports test results for distributional differences between chaebol and control firms. Control firms are obtained using the matching estimator technique that selects firms with the shortest Euclidean distances from chaebol firms in the space of covariates as matching firms. Control firms are much more similar to chaebol firms than independent firms even though the distribution of three variables – Tobin’s Q, firm size, and liquidity – are significantly different between chaebol and control firms. Even for these variables, the differences of the 25, 50, and 75th values are much smaller than those of Panel A. Therefore, the control group of firms, not just non-chaebol independent firms, are better candidates for comparison with chaebol firms.
4. Valuation Implication of Group Affiliation
Table 3 shows the average treatment effects on the treated (ATT), i.e. the effects of chaebol affiliation on firm value, from 1990 to 2010. Control firms are selected by using the matching estimator technique based on the values of covariates as at 1990. ATT indicate that firm value as measured by Tobin’s Q increases by affiliation to business groups.2 When control firms are matched to chaebol firms in 1990 and over 4 years afterwards, there are no value consequences of business group affiliation. Since we compare two groups of firms with almost identical firm characteristics, this result is not totally unexpected. From 2001, however, chaebol-affiliated firms have higher values than control firms with similar firm characteristics without any exception until 2010. Since we compare chaebol and non-chaebol firms after controlling for observables, this result suggests that business group affiliation has a value implication. In other words, business group structure seems to have a positive impact on firm value, other things being equal.3
| Year | Q | Coeff. | SE | z-value | p > |z| |
|---|---|---|---|---|---|
| 1990 | ATT | 0.0202 | 0.0307 | 0.66 | 0.511 |
| 1991 | ATT | −0.0013 | 0.0270 | −0.05 | 0.960 |
| 1992 | ATT | 0.0089 | 0.0309 | 0.29 | 0.774 |
| 1993 | ATT | 0.1259 | 0.0782 | 1.61 | 0.107 |
| 1994 | ATT | 0.0782 | 0.0554 | 1.41 | 0.158 |
| 1995 | ATT | 0.1081 | 0.0420 | 2.58*** | 0.010 |
| 1996 | ATT | 0.0866 | 0.0351 | 2.47** | 0.014 |
| 1997 | ATT | 0.0344 | 0.0341 | 1.01 | 0.314 |
| 1998 | ATT | 0.0158 | 0.0508 | 0.31 | 0.756 |
| 1999 | ATT | 0.1834 | 0.1217 | 1.51 | 0.132 |
| 2000 | ATT | 0.2698 | 0.1941 | 1.39 | 0.165 |
| 2001 | ATT | 0.4213 | 0.2349 | 1.79* | 0.073 |
| 2002 | ATT | 0.1713 | 0.0614 | 2.79*** | 0.005 |
| 2003 | ATT | 0.1447 | 0.0502 | 2.88*** | 0.004 |
| 2004 | ATT | 0.1050 | 0.0468 | 2.25** | 0.013 |
| 2005 | ATT | 0.2363 | 0.0698 | 3.39*** | 0.001 |
| 2006 | ATT | 0.1771 | 0.0635 | 2.79*** | 0.005 |
| 2007 | ATT | 0.2976 | 0.0763 | 3.90*** | 0.000 |
| 2008 | ATT | 0.1048 | 0.0631 | 1.66* | 0.097 |
| 2009 | ATT | 0.2387 | 0.0618 | 3.87*** | 0.001 |
| 2010 | ATT | 0.2511 | 0.0655 | 3.83*** | 0.000 |
The advantage of group affiliation tends to shrink over economic crisis periods. For example, the statistically significant firm value advantage in 1995 and 1996 disappears during the 1997 Asian currency crisis and the periods following. The firm value advantage of chaebol affiliation is 0.1048 in 2008 when the global economic crisis occurs, which is much smaller in its magnitude and statistical significance compared to other years. During times of economic hardship all firms suffer from slow business activity and credit crunches, which leads to declines in stock price for both chaebol and control firms, and the value difference therefore becomes narrower.
Figure 1 shows the changes in average stock prices for chaebol group firms and control group firms over time. The differences are not significant during the 1990s, but the gap between average prices of the two groups diverges significantly during the 2000s, suggesting the valuation advantage of group affiliation.
Changes in average stock prices of chaebol and control firms. This figure illustrates the changes in average stock prices for chaebol and control firms from 1990 to 2010. Average stock prices are obtained by calculating the arithmetic mean of mean prices of firms included in chaebol and control groups for each fiscal year. The mean price of a firm is the average of daily closing prices for a year.
| Mean | SE | SD | |
|---|---|---|---|
| Chaebol firms | 5.4506 | 1.2267 | 11.0399 |
| Control firms | 1.8558 | 0.6966 | 4.8760 |
| Difference | 3.5948** | t-value = 2.1533 | |
5. Determinants of Performance
We have shown that chaebol firms have higher firm values, even though we compare chaebol firms with control firms that have similar firm characteristics and are from the same industry. We now analyze why chaebol firms come to have higher values. We conjecture that this is achieved by changes in firm characteristics, not simply by business group affiliation. Figure 2 illustrates changes in covariate values over time.4 Although our matching estimator technique generates control firms that are very similar to chaebol firms as at 1990, covariate values change over time. If covariate values become significantly different between chaebol and control firms as in foreign shareholdings in Figure 2, the differences in firm characteristics will cause differential changes in firm value over time.
Changes in covariate values of chaebol and control firms over time. This figure illustrates the changes in covariate values over time. Control firms are selected based on covariate values as of 1990. The total sample period runs from 1981 to 2010.
Table 5 shows means and medians of variables that are known to be related to firm value in the existing literature for chaebol and control firms. We know from Table 1 that there are no differences in values of profitability, asset growth rate, leverage, largest shareholdings, and foreign ownership between chaebol and control firms as at 1990. However, these characteristics may become different between the two groups as time passes, and these differences may cause differential firm values.
| Mean | Median | |||||||
|---|---|---|---|---|---|---|---|---|
| Chaebol | Control | Difference | t-value | Chaebol | Control | Difference | Pearson χ2 | |
| Q | 1.011 | 0.927 | 0.084*** | 3.38 | 0.952 | 0.847 | 0.105*** | 93.74 |
| Size | 20.80 | 19.93 | 0.863*** | 20.07 | 20.79 | 20.01 | 0.78*** | 228.89 |
| Profitability | 0.059 | 0.052 | 0.007** | 2.55 | 0.056 | 0.056 | 0.000 | 0.005 |
| Growth | 0.099 | 0.069 | 0.030*** | 3.91 | 0.090 | 0.064 | 0.026*** | 26.26 |
| Investment | 0.073 | 0.052 | 0.021*** | 5.20 | 0.063 | 0.047 | 0.016*** | 31.92 |
| R&D | 0.085 | 0.015 | 0.070*** | 6.98 | 0.000 | 0.000 | 0.000 | 0.006 |
| Debt | 0.635 | 0.640 | −0.005 | −0.19 | 0.644 | 0.623 | 0.021*** | 11.91 |
| Liquidity | 1.241 | 1.267 | −0.026 | −0.79 | 1.022 | 1.109 | −0.087*** | 16.29 |
| Dividend | 0.294 | 0.337 | −0.044 | −1.43 | 0.187 | 0.191 | −0.004 | 0.061 |
| Big | 0.274 | 0.280 | −0.006 | −1.06 | 0.240 | 0.268 | −0.028*** | 10.23 |
| Foreign | 0.115 | 0.084 | 0.030*** | 6.45 | 0.057 | 0.035 | 0.022*** | 30.95 |
The results in Table 5 indicate that all except the dividend variable are significantly different between chaebol and control firms based on either the mean test or the median test for the whole period. According to the mean tests, size, profitability, growth, investment, R&D, and foreign variables are significantly different between the two groups. The results from the median tests indicate that corporate leverage, liquidity, and equity ownership of the largest shareholder are also significantly different between chaebol and control firms. These differences in firm characteristics may have caused differences in Tobin’s Q.
Table 6 provides more detailed information in this regard. It shows the mean differences of each variable between treated, i.e. chaebol firms, and corresponding control firms for the three sub-periods. The Asian financial crisis had an unprecedented impact on the Korean economy and many firms went through drastic restructuring in order to survive. It is generally believed that business practices and firm characteristics are very different between pre-crisis and post-crisis periods. Therefore, the whole sample period is split into three sub-periods: pre-crisis period (1990–1996), crisis and aftermath period (1997–2002), and post-crisis period (2003–2010). Figure 3 illustrates changes in the KOSPI after the beginning of our sample period. Our second sub-period covers the period of the Asian financial crisis and its aftermath before the KOSPI begins to move up in 2003.5
| (1) 1990–1996 | (2) 1997–2002 | (3) 2003–2010 | (2)–(1) | (3)–(2) | (3)–(1) | |
|---|---|---|---|---|---|---|
| Q | ||||||
| Treated | 1.073 | 0.880 | 1.055 | −0.193***(−7.77) | 0.176*** (7.70) | −0.018 (−0.69) |
| Control | 0.995 | 0.932 | 0.864 | −.063 (−0.89) | −0.068 (−0.99) | −0.131*** (−7.19) |
| Diff. t-value | 0.078*** (3.65) | −0.052 (0.68) | 0.191*** (8.29) | −0.130* (−1.74) | 0.243*** (3.38) | 0.113*** (3.56) |
| Size | ||||||
| Treated | 20.155 | 20.946 | 21.242 | 0.791*** (10.29) | 0.296*** (3.54) | 1.087*** (14.03) |
| Control | 19.536 | 20.097 | 20.156 | 0.561*** (8.59) | 0.059 (0.93) | 0.619*** (10.72) |
| Diff. t-value | 0.618*** (9.42) | 0.849*** (10.95) | 1.086*** (15.56) | 0.230** (2.28) | 0.238** (2.26) | 0.468*** (4.84) |
| Profitability | ||||||
| Treated | 0.059 | 0.061 | 0.058 | 0.002 (0.26) | −0.003 (−0.83) | −0.002 (−0.30) |
| Control | 0.063 | 0.050 | 0.044 | −0.014*** (3.53) | −0.006 (1.29) | −0.019*** (−6.62) |
| Diff. t-value | −0.004 (−0.77) | 0.011** (2.23) | 0.013*** (3.79) | 0.016** (2.04) | 0.002 (.37) | 0.018*** (2.77) |
| Growth | ||||||
| Treated | 0.158 | 0.058 | 0.077 | −0.100*** (−7.45) | 0.019 (1.21) | −0.081*** (−6.14) |
| Control | 0.155 | 0.017 | 0.032 | −0.138*** (−12.37) | 0.016 (1.24) | −0.122*** (−12.35) |
| Diff. t-value | 0.003 (0.32) | 0.041*** (2.65) | 0.045*** (3.41) | 0.038** (2.19) | 0.004 (0.18) | 0.042** (2.54) |
| Investment | ||||||
| Treated | 0.116 | 0.052 | 0.052 | −0.064*** (−10.37) | 0.001 (0.07) | −0.063*** (−9.57) |
| Control | 0.102 | 0.024 | 0.030 | −0.078*** (−10.86) | 0.006 (0.79) | −0.072*** (−13.29) |
| Diff. t-value | 0.013** (2.50) | 0.028*** (3.40) | 0.022*** (3.42) | 0.015 (1.56) | −0.006 (−0.53) | 0.009 (1.08) |
| R&D † | ||||||
| Treated | – | 0.074† | 0.168 | – | 0.094*** (3.13) | – |
| Control | – | 0.013 | 0.030 | – | 0.017*** (4.24) | – |
| Diff. t-value | – | 0.061*** (3.57) | 0.138*** (6.11) | – | 0.077** (2.55) | – |
| Debt | ||||||
| Treated | 0.759 | 0.671 | 0.501 | −0.088*** (−3.69) | −0.171*** (−11.05) | −0.259*** (−11.89) |
| Control | 0.708 | 0.782 | 0.475 | 0.074 (1.07) | −0.306*** (−4.72) | −0.232*** (−25.68) |
| Diff. t-value | 0.052** (2.51) | −0.111 (−1.46) | 0.025** (2.00) | −0.162** (−2.22) | 0.136** (2.04) | −0.027 (−1.13) |
| Liquidity | ||||||
| Treated | 1.110 | 1.115 | 1.450 | 0.005 (0.10) | 0.335*** (4.79) | 0.340*** (5.94) |
| Control | 1.152 | 1.160 | 1.450 | 0.008 (0.21) | 0.290*** (4.54) | 0.298*** (5.24) |
| Diff. t-value | −0.042 (−1.36) | −0.044 (−0.81) | 0.000 (0.00) | −0.003 (−0.04) | 0.045 (0.47) | 0.042 (0.52) |
| Dividend | ||||||
| Treated | 0.411 | 0.258 | 0.208 | −0.153*** (−3.57) | −0.050 (−1.20) | −0.203*** (−7.45) |
| Control | 0.531 | 0.206 | 0.255 | −0.325*** (−5.10) | 0.050 (.93) | −0.276*** (−3.84) |
| Diff. t-value | −0.121** (−2.03) | 0.052 (1.15) | −0.048 (−0.97) | 0.172** (2.24) | −0.099 (−1.47) | 0.073 (.95) |
| Big | ||||||
| Treated | 0.216 | 0.284 | 0.324 | 0.068*** (6.68) | 0.040*** (3.10) | 0.108*** (10.48) |
| Control | 0.231 | 0.284 | 0.327 | 0.052*** (6.50) | 0.043*** (4.44) | 0.095*** (11.41) |
| Diff. t-value | −0.015** (−2.13) | 0.001 (0.07) | −0.003 (−0.23) | 0.016 (1.23) | −0.003 (−0.21) | 0.013 (0.95) |
| Foreign | ||||||
| Treated | 0.062 | 0.111 | 0.171 | 0.049*** (7.12) | 0.060*** (6.11) | 0.109*** (14.07) |
| Control | 0.052 | 0.059 | 0.137 | 0.007 (1.33) | 0.078*** (8.88) | 0.085*** (11.59) |
| Diff. t-value | 0.010** (2.50) | 0.052*** (6.31) | 0.034*** (3.42) | 0.042*** (4.86) | −0.018 (−1.36) | 0.024** (2.27) |
Changes in the Korea Composite Stock Price Index (KOSPI). This figure shows the historic performance of the KOSPI after 1990. There are two sharp declines: one is the Asian financial crisis of 1997 and the other is the global economic crisis represented by the collapse of Lehman Brothers, Inc. in 2008.
Table 6 shows how firm value and firm characteristics change over different sub-periods for chaebol and control firms. It also reports difference-in-difference estimators that are changes in differences between chaebol and control firms over two different sub-periods after controlling for unobservables.
5.1. Firm Value
The average value of Tobin’s Q is larger for chaebol firms than control firms before the Asian financial crisis. During the crisis and after, firm values decrease for both chaebol-affiliated and control firms, but only chaebol firms show a statistically significant drop (−0.193) in value. After the intermediate sub-period the Tobin’s Q of chaebol firms increases by 0.176, which is very significant statistically, but control firms show an insignificant decrease in Tobin’s Q, which results in much bigger Tobin’s Q for chaebol firms than control firms during the post-crisis period. The difference in Tobin’s Q between pre- and post-crisis periods for chaebol firms is not statistically different from zero, but that for control firms is −0.131, which is statistically significant. These results suggest that chaebol firms successfully adapt to new business environments after a crisis, while control firms fail. Similar conclusions can be derived from difference-in-difference estimators. The difference in Tobin’s Q between chaebol and control firms decreases from the pre-crisis period to the crisis period by −0.130. After that, however, the difference in firm value widens significantly from the second sub-period to the post-crisis period. The difference-in-difference estimator between pre- and post-crisis period is 0.113, which implies that the firm value advantage of chaebol firms is strengthened over time.
5.2. Size
Firm size is an important determinant of firm value. Larger firms suffer less from information asymmetry, which will have a positive impact on firm value. However, larger firms cannot adapt to ever-changing business environments as effectively as smaller firms. The extant empirical research is also inconclusive on the effect of firm size on firm value (Bhattacharyya and Arunima, 2009). Table 6 shows that chaebol firms are, on average, larger than control firms, which is true for all sub-periods. The results on mean differences between the two sub-periods suggest that firm size increases over time for both chaebol and control firms. Difference-in-difference estimators are all positive and statistically significant, which indicates that the size difference between chaebol and control firms widens over time. It was possible to increase firm size without positive economic consequences through an inefficient financial system directed by the government before the Asian financial crisis. However, this undeserved expansion of business became impossible after the crisis since the IMF required the Korean government to apply very rigorous guidelines to market participants. Therefore, a higher increase in firm size of chaebol firms over control firms even after the financial crisis is consistent with the results of Tobin’s Q in the sense that only firms with economic success can grow quickly after a crisis.
5.3. Profitability
The profitability of chaebol and control firms is not different before the crisis. However, chaebol firms are more profitable than control firms over the second and third sub-periods. As for the change in profitability, chaebol firms do not show any significant increase or decrease in profitability over time. However, control firms’ profitability decreases significantly from the pre-crisis period to the crisis and aftermath period. Difference-in-difference estimators indicate that the difference in profitability between chaebol and control firms increases from the pre-crisis period to the crisis period. The worsened profitability of control firms is not recovered during the post-crisis period, which leads to a significant increase in the profitability gap between chaebol and control firms over the entire sample period, which is equivalent to 1.8% of the total asset value. Obviously, it is possible that the higher profitability of chaebol firms is one of the major factors that contributes to the higher firm value of chaebol-affiliated companies.
5.4. Growth
Growth rate is another important factor that has an impact on firm value. During the pre-crisis period there is no difference in growth rate between chaebol and control firms. However, during the crisis period and the post-crisis period chaebol firms show considerably higher growth rates than control firms, and the differences in growth rates are 4.1% and 4.5%, respectively. Among the three sub-periods the pre-crisis period shows the highest growth rate, which is followed by the lowest growth rate during the crisis period. The growth rate of the post-crisis period is higher than that of the crisis period, but the difference is not statistically significant for both chaebol and control firms. Difference-in-difference estimators are 3.8% and 4.2% for the comparison of first and second sub-periods and first and third sub-periods, respectively. This result implies that the growth rate of chaebol firms becomes higher than control firms as time passes. The higher growth rate of chaebol firms may therefore explain the higher firm value of chaebol companies.
5.5. Investments
As proxies for investments, this paper employs cash flow from investment activities and research and development (R&D) expenses. For all sub-periods investments of chaebol firms are higher than control firms. Investment activities during the pre-crisis period are the highest among the three sub-periods for both chaebol and control firms and they decrease sharply during the crisis period. The decreased investments are stable during the post-crisis period. Difference-in-difference results suggest that the gap in investment activities between chaebol and control firms does not change significantly over time.
The R&D variable has missing values for the pre-crisis period. For the other two sub-periods chaebol firms’R&D expenses are higher than those of control firms, which is consistent with the results of the investment variable. As shown in the comparison of the second and third sub-periods, both chaebol and control firms increase R&D expenses after the Asian financial crisis. Note that the difference-in-difference estimator, 0.077%, is significantly positive, which is obtained by a 0.094% increase in R&D expenses by chaebol firms and only a 0.017% increase by control firms. The big difference in R&D expenses between chaebol and control firms will cause a significant firm value difference via performance improvement.
5.6. Leverage
Corporate leverage has been the most frequently mentioned variable in discussions of firm value following the seminal paper by Modigliani and Miller (1958). Leverage can have a positive impact on firm value through the tax shield effect and monitoring by capital markets. Excessive debt can also decrease firm value via increased risk of bankruptcy and agency costs such as overinvestment and underinvestment. Table 6 shows the debt ratios of chaebol and control firms for the three sub-periods. Except for the crisis period chaebol firms employ higher debt than control firms. However, the debt level of Korean firms has decreased continuously regardless of chaebol affiliation. Before the Asian financial crisis companies had been criticized for having excessive debt that triggered the crisis. The Korean government, following a strong recommendation by the IMF, urged companies, especially chaebol-affiliated companies, to decrease their debt level. This policy mandate seems to cause a negative difference-in-difference estimator of −16.2% in the comparison of pre-crisis and crisis periods, wherein chaebol firms decrease their debt level significantly but control firms do not. However, control firms sharply decrease their leverage afterwards both voluntarily to survive in a new business environment and involuntarily on the direction of the government. Obviously, the difference in corporate leverage between chaebol and control firms is one of the major determinants of firm value.
5.7. Liquidity
Many papers have studied the effect of liquidity on firm value (Acharya et al., 2007; Almeida and Campello, 2007). If capital markets are perfect, companies can raise the required capital for their investments by paying the appropriate costs of capital and they do not have to maintain liquidity for their investments. If capital markets are imperfect, however, a variety of market frictions make it difficult for firms to invest even on positive NPV projects. Therefore, liquidity can have a positive influence on firm value in practice. Table 6 shows that there are no differences in liquidity between chaebol and control firms. Note that companies sharply increase their liquidity irrespective of chaebol affiliation during the post-crisis period, which implies that the value of liquidity increases after crisis.
5.8. Dividends
There have been three positions regarding the relationship between dividend payments and firm value: dividend irrelevancy and positive and negative relationships. Dividend controversy is still inconclusive and the effect of dividends on firm value is an empirical issue. Chaebol firms pay lower dividends than control firms before the crisis, which is consistent with the evidence of higher investments of chaebol firms in the pre-crisis period. During the crisis period firms decrease dividends sharply due to the credit crunch, and control firms decrease dividends significantly more than chaebol firms. After the Asian financial crisis there is no difference between chaebol and control firms in dividend payments.
5.9. Governance
Corporate governance has been cited as an important factor that has an impact on firm value in the extant literature. Ownerships of the biggest shareholders and foreigners are used in this paper as proxies for corporate governance. Conventional agency theory predicts a positive relationship between share ownership of largest shareholders and firm value. Foreigners are believed to be in a better position to monitor the entrenchment of owner–managers, which means there is a positive relationship between foreign ownership and firm value. As shown in Table 6, the ownership of the largest shareholders for chaebol-affiliated firms is smaller than control firms before the crisis. Both chaebol and control firms have continuously increased the share ownership of the largest shareholders over time. Currently, there is no difference in the largest shareholdings between chaebol and control firms.
As for foreign ownership, chaebol firms have significantly higher foreign ownership than control firms for all sub-periods. Foreigners are, in general, known to invest in well-known, bigger companies. The results in Table 6 indicate that another important consideration for them is chaebol affiliation since foreign ownership is much higher for chaebol firms. Foreign ownership has continuously increased over our sample period. This can partly be explained by the liberalization of capital markets that accelerated after the Asian financial crisis. Difference-in-difference estimators are significantly positive when comparing the pre-crisis period with the other two sub-periods. This suggests that the gap in foreign ownership between chaebol and control firms has widened. Currently, foreigners invest much more in chaebol-affiliated firms than control firms.
Overall, chaebol firms become larger and more profitable, grow faster with more investments, maintain a higher debt level, and are owned by more foreign investors. A higher profitability and more investments can make chaebol firms grow faster, resulting in bigger firm size (Kim, 2011). A higher leverage can contribute to firm value through a larger tax shield and a more active monitoring by capital markets. Shareholder activism by foreign shareholders can also prevent entrenchment by owner–managers, which has a positive impact on the value of chaebol firms (Bae et al., 2002).
6. Difference in Firm Value
In order to determine the effects of covariates on firm value, we perform panel data analysis where Tobin’s Q is a dependent variable. In all models industry dummies are included to control the possibility that Tobin’s Qs are systematically different from industry to industry. Table 7 shows the results. Over an entire period, the chaebol dummy shows a significantly positive coefficient, indicating that the firm value of chaebol-affiliated companies is higher than control firms. Variables that have significantly positive impacts on firm value are investment, R&D, debt, liquidity, big, and foreign. Therefore, firms with more investments, with a higher leverage and liquidity, and with higher largest shareholding and foreign shareholding have higher firm value. But firm size turns out to be a negative factor for firm value.
| Q | Total period (1990–1996) | Pre-crisis (1990–1996) | Crisis and aftermath (1997–2002) | Post-crisis (2003–2010) | ||||
|---|---|---|---|---|---|---|---|---|
| Coef. | z | Coef. | z | Coef. | z | Coef. | z | |
| Chaebol | 0.087*** | 2.54 | 0.066** | 2.13 | 0.039 | 1.42 | 0.125* | 1.91 |
| Size | −0.019*** | −2.46 | −0.042*** | −4.30 | −0.013 | −1.13 | 0.046** | 2.23 |
| Profitability | 0.339 | 3.29 | 0.194 | 1.24 | 0.039 | 0.26 | 0.552*** | 2.77 |
| Growth | 0.033 | 1.32 | −0.057* | −1.78 | −0.038 | −1.14 | 0.033 | 0.73 |
| Investment | 0.303*** | 5.63 | −0.041 | −0.64 | 0.353*** | 3.95 | 0.156* | 1.86 |
| R&D | 11.543*** | 5.16 | 17.625*** | 4.88 | 3.473 | 0.64 | ||
| Debt | 0.424*** | 10.61 | 0.443*** | 6.66 | 0.536*** | 9.94 | 0.342*** | 2.90 |
| Liquidity | −0.008 | −1.00 | 0.054*** | 3.66 | −0.013 | −1.21 | 0.002 | 0.11 |
| Dividend | 0.037 | 1.58 | 0.016 | 0.77 | −0.071* | −1.90 | −0.213*** | −3.33 |
| Big | 0.035 | 0.97 | 0.134* | 1.94 | 0.013 | 0.28 | 0.118* | 1.73 |
| Foreign | 0.686*** | 12.52 | 0.507*** | 5.26 | 0.447*** | 5.59 | 0.479*** | 4.28 |
| Industry Dummy | Included | Included | Included | Included | ||||
| Cons | 1.041*** | 3.46 | −0.252 | −0.49 | ||||
| R 2 | 0.3171 | 0.3918 | 0.5573 | 0.4460 | ||||
Firms that invest more and have more R&D expenses have a potential for better performance, which can lead to higher firm value, ceteris paribus. Theoretically, corporate leverage has both positive and negative impacts on firm value. Too much debt above optimal capital structure can generate higher expected bankruptcy and agency costs. On the other hand, debt can reduce free cash flows and lead to more effective monitoring by capital markets (Jensen, 1986). The significantly positive coefficient on the debt variable suggests that the latter effects are stronger in the case of Korean firms. Corporate liquidity can free firms of unnecessary concern for technical default and prepare companies for capital needs caused by sudden investment opportunities. Our empirical result on the liquidity variable is consistent with this argument.
Many studies assert that a disparity between cash flow and voting rights negatively impacts firm value (Joh, 2003; Baek et al., 2004). The traditional agency cost argument also predicts that firms with partial ownership will incur agency costs that similar firms with 100% ownership do not incur. A positive sign on the coefficient for the largest shareholding is consistent with these theories. Finally, foreigners will have a positive impact on firm value through more effective monitoring of company activities. Baek et al. (2004) find that during the Asian financial crisis, firms with higher ownership concentration by unaffiliated foreign investors show a smaller reduction in share value. The positive coefficient for the foreign ownership variable is consistent with their result.
As for the sub-period analysis, coefficients on the chaebol dummy variable are all positive with a statistical significance over pre- and post-crisis periods. The coefficient value of 0.119 for the analysis of the post-crisis period is much larger than the other two sub-periods. Since the equilibrium value of Tobin’s Q is equal to 1, chaebol affiliation can produce value that is 11.9% higher than otherwise equivalent non-chaebol firms.
The size variable is a negative factor for firm value during the pre-crisis and crisis periods. That is, other things being equal, bigger companies have a lower firm value. However, the coefficient on the size variable is significantly positive over the post-crisis period. Many scholars agree that the Asian financial crisis provided a chance to revamp the Korean economy. Before the crisis firms attempted to increase their size at the expense of efficiency, since size alone has various advantages. For example, to develop the economy faster, the Korean government provided capital to the 30 largest business groups at below-market costs of capital. However, since the capital was raised at cheaper than the market rate, firms tended to waste the capital in business areas where they did not have any expertise. This corporate behavior led to lower firm value for bigger companies.
However, the corporate landscape has been changed dramatically by the Asian financial crisis. Firms have struggled to survive and have had to improve the efficiency of their business operations by drastically downsizing and restructuring. Now firm size can be a positive factor for firm value since the negative effects of firm size have been eliminated, while the positive effects such as economies of scale and less information asymmetry remain the same. The significantly positive coefficient on the size variable is consistent with this argument. The efficiency argument is also supported by the result on the profitability variable. During the previous two sub-periods profitability was an insignificant factor for increase in firm value. However, under a tighter and more efficient economic system, more profitable firms generate higher firm value.
7. Relative Importance of Covariates
In this subsection we will investigate the relative importance of covariates in explaining the difference in firm value between chaebol and control firms. Table 6 shows the differences in covariate values between chaebol and control firms. Table 8 presents the contribution each covariate makes to the value of Tobin’s Q based on the results from firm value regression in Table 7. We consider the results of the post-crisis period since they reflect the situation most similar to the current time. Column (1) shows coefficients from firm value regression over the post-crisis period in Table 7. Column (2) presents differences in covariate values between chaebol and control firms over the post-crisis period shown in Table 6. By multiplying columns (1) and (2), we can calculate the change in Tobin’s Q that is caused by the differences in covariate values. For the post-crisis period, covariate differences between chaebol and control firms increase Tobin’s Q by .0915, which is equivalent to 9.15% of equilibrium Tobin’s Q value of 1. The last column expresses the contribution of each covariate in percentage terms.
| (1) | (2) | (1) * (2) | Proportions explained by each firm characteristic (%) | |
|---|---|---|---|---|
| Size | 0.044 | 1.086 | 0.0478 | 52.2 |
| Profitability | 0.508 | 0.013 | 0.0066 | 7.2 |
| Growth | 0.021 | 0.045 | 0.0009 | 1.0 |
| Investment | 0.168 | 0.022 | 0.0037 | 4.0 |
| R&D | 3.866 | 0.001 | 0.0053 | 5.8 |
| Debt | 0.396 | 0.025 | 0.0099 | 10.8 |
| Liquidity | 0.007 | 0.000 | 0.0000 | 0.0 |
| Dividend | −0.004 | −0.048 | 0.0002 | 0.2 |
| Big | 0.146 | −0.003 | −0.0004 | −0.5 |
| Foreign | 0.513 | 0.034 | 0.0174 | 19.1 |
| Total | 0.0915 | 100.0 |
Among the many factors, firm size is the most important, explaining more than 50% of contributions. Foreign ownership is the next important determinant of firm value accounting for 19.1% of an increase in Tobin’s Q. Leverage is also important for firm value, explaining 10.8% of the increase in Tobin’s Q. Other important factors are profitability (7.2%), R&D expenses (5.8%), and investments (4.0%). Companies have to pay closer attention to these factors, i.e. firm size, foreign ownership, leverage, profitability and investment, to improve firm value. Note that foreign ownership is a major determinant of firm value increase that contributes more than leverage and profitability, suggesting the importance of a better governance structure.
8. Analysis Using Difference Variables
Table 9 presents regression results where we use changes in value for each variable over pre-crisis, crisis and aftermath, and post-crisis periods. For example, dQ for the pre-crisis period is obtained by subtracting the value of Q each year from the value of Q at the end of sub-period (1996). Industry dummies are included to control the possibility that Tobin’s Q can be systematically different among different industries. Also, heteroscedasticity-consistent standard errors are used to calculate z-values. By using difference variables we can infer the marginal effect of the change in each variable to the change in firm value.
| dQ | Pre-crisis | Crisis and aftermath | Post-crisis | |||
|---|---|---|---|---|---|---|
| Coef. | z-value | Coef. | z-value | Coef. | z-value | |
| Chaebol | −0.078** | −2.45 | 0.009 | 0.29 | −0.060 | −1.17 |
| dSize | −0.028* | −1.84 | 0.056 | 0.88 | 0.114** | 1.98 |
| dProfitability | −0.030 | −0.42 | 0.396*** | 3.69 | 0.682** | 2.35 |
| dGrowth | −0.047 | −1.37 | 0.006 | 0.22 | −0.018 | −0.24 |
| dInvestment | 0.046 | 0.68 | −0.184*** | −2.62 | 0.186 | 1.35 |
| dR&D | – | – | 14.561* | 1.83 | −5.171 | −0.57 |
| dDebt | 0.979*** | 31.3 | 1.014*** | 108.53 | 0.203 | 0.88 |
| dLiquidity | 0.061*** | 4.17 | 0.007 | 0.99 | 0.010 | 0.34 |
| dDividend | −0.002 | −0.37 | 0.001 | 0.15 | −0.008 | −0.74 |
| dBig | 0.161* | 1.93 | 0.017 | 0.31 | 0.178** | 2.14 |
| dForeign | 0.805*** | 6.99 | 0.401*** | 3.34 | 0.595*** | 3.08 |
| Industry dummy | Included | Included | Included | Included | ||
| Cons | 0.068 | 0.53 | −0.061 | −1.64 | 0.081 | 0.89 |
| R 2 | 0.7267 | 0.9727 | 0.3160† | |||
The results are consistent with those reported in Table 7. Consistently significant variables across more than two sub-periods are increases in profitability and leverage, and increases in largest shareholder ownership and foreign ownership. Other things being equal, an increase in profitability will result in higher firm value. An increase in the largest shareholder ownership will contribute to firm value increase through a reduction of agency costs. A positive relationship between changes in leverage and foreign ownership and a change in firm value suggests that monitoring by capital markets and foreign investors is very important in enhancing firm value. An increase in firm size has a negative relationship with an increase in firm value during thr pre-crisis period, but it has a significantly positive relationship with an increase in firm value during the post-crisis period. This result is consistent with the results found in Table 7. Due to the Asian financial crisis Korean firms improved operational efficiency via downsizing and restructuring, which eliminated the negative effects of firm size. The results in Table 9 indicate that firm size is no longer a negative factor for firm value after the crisis.
Finally, the coefficient for the chaebol dummy is significantly negative before the crisis. However, it becomes insignificant during and after the crisis. This result suggests that changes in firm value are explained by changes in covariates, not by chaebol affiliation. In other words, chaebol affiliation per se is not a factor that induces value increase. Rather, we observe changes in value occurring due to changes in covariates in the long term. The main driving forces for higher firm value of chaebol companies over control companies in the long term are changes in firm characteristics, not simply chaebol affiliation.
9. Conclusions
In this paper we investigated the long-term value implication of business group affiliation. In order to perform a valid comparison between business-group-affiliated firms and independent firms, an appropriate benchmark must be formed. For this we employed a matching estimator technique that forms a control group by selecting firms with characteristics most similar to chaebol firms.
The advantage in firm value of chaebol affiliation is confirmed by stock price performance in the capital market. We conjecture that this difference in firm values between chaebol and control firms is caused by differences in firm characteristics. In fact, we find that firm characteristics such as size, profitability, asset growth, investments, leverage, liquidity, and ownership structure become significantly different between chaebol and control firms as time passes, although they were very similar at the start of our sample period. Difference-in-difference estimators allow us to determine the time series pattern of differences in firm characteristics between chaebol and control firms. Compared with control firms, chaebol firms, over time, become bigger and more profitable, grow faster, invest more, have higher debt, and have higher foreign ownership. The result suggests that higher profitability and more investments enable chaebol firms to grow faster, resulting in a bigger firm size. Chaebol firms get benefits from a higher tax shield and more active monitoring by banks and capital markets due to a higher leverage. Increased foreign ownership may have also contributed to the increase in firm value for chaebol-affiliated companies by checking the entrenchment of owner–managers.
Firm value regressions over the sub-periods show that the chaebol dummy variable has a positive sign for all three sub-periods with statistical significance for the pre-crisis and post-crisis periods. It indicates that chaebol affiliation has a positive impact on firm value. Firm characteristics that contribute to firm value are firm size, profitability, investments, research expenses, leverage, dividends, largest shareholding, and foreign ownership. Among these, profitability, investments, research expenses, leverage, largest shareholding, and foreign ownership have a positive impact on firm value, and dividend payments have a negative impact on firm value. Firm size has significantly negative coefficients over th pre-crisis and crisis periods, but has a significantly positive sign over the post-crisis period. In terms of strength of influence on firm value, size is the most influential factor followed by foreign ownership, leverage, profitability, and investments. The fact that foreign ownership is the second important determinant of firm value suggests that the influence of foreign investors has increased significantly in the Korean economy.
When we regress changes in firm value on difference variables, we obtain similar results as in the previous analysis. Changes in profitability, leverage, ownership of the largest shareholder, and foreign ownership have a positive relationship with an increase in firm value. Increase in size is a significantly negative factor for firm value increase over the pre-crisis period, but it becomes a significantly positive factor after the crisis. The coefficient on the chaebol dummy is significantly negative before the crisis, but it is not significant at all over the crisis and post-crisis periods. This result implies that the change in firm value is achieved by changes in firm characteristics, not by chaebol affiliation per se. In other words, chaebol firms have a higher firm value than control firms in the long term, not because they are affiliated with business groups, but because their firm characteristics have changed to produce a higher firm value.
Given that firm value changes are explained by changes in firm characteristics, not by chaebol affiliation, an interesting future research agenda is to investigate why chaebol firms, compared to control firms, come to have such firm characteristics. As we demonstrate, two groups of firms diverge in terms of covariate values even though they start as similar firms. If there are some elements that induce systematic differences between the two groups, they are good subjects for future research. Possible candidates are a more efficient resource allocation through internal capital markets, an economic environment that favors business group structure, the cozy relations between politics and business, and so forth. Future research on these issues can provide good guidelines for economic policy on business groups.
