Commodity price hedging: A chaebol perspective

(1)

by

Nils de Jong

University of Groningen &

Uppsala University

Faculty of Economics and Business

MSc. International Financial Management

(2)

Commodity price hedging: A chaebol perspective

By Nils de Jong

Abstract

The aim of this master thesis is to examine the effect of firm characteristics – debt-equity ratio, dividend pay-out ratio, quick ratio, capital expenditures, and firm size - on commodity price hedging and whether hedging incentives are affected by industrial grouping membership. Using a two-step regression model for a sample of South Korean firms in the 1994-2001 period, I find evidence that member-firms, chaebol, have less incentive to hedge than independent firms. I also find that the industrial grouping membership effect weakens over time as chaebol-affiliated firms are more likely to hedge after the Asian financial crisis due to tighter financial constraints. In both the pre-crisis and post-crisis period, non-chaebol firms with a low liquidity position or high leverage are more likely to hedge.

Keywords: Chaebol, commodity, conglomerate, exposure, hedging, South Korea

Research theme: Commodity price hedging

(3)

2.5 Conclusion ... 17 3. Methodology ... 19 3.1 Model description ... 19 3.2 Operationalization variables ... 21 3.2.1 First-step regression... 21 3.2.2 Second-step regression ... 22 4. Data description ... 26 4.1 Selection procedure ... 26 4.2 Sampling period ... 27 4.3 Descriptive Statistics ... 27

4.3.1 Summary statistics commodities ... 27

4.3.2 Descriptive statistics South Korean firms ... 28

4.4 Reliability and correctness model ... 29

5. Empirical results first-step regression ... 33

(4)

5.2 Commodity price exposure: Before and after the 1997 Asian crisis ... 35

5.3 Commodity price exposure: chaebol versus non-chaebol ... 38

6. Determinants of the commodity price exposure ... 40

6.1 Financial distress ... 40

6.1.1 Leverage ... 40

6.1.2 Liquidity position ... 41

6.2 Investment opportunities ... 42

6.3 Firm size ... 43

7. Empirical results second-step regression ... 45

7.1 Industrial grouping effect ... 46

7.2 Hedging practices before and after corporate reforms ... 47

7.2.1 Leverage ... 47

7.2.2 Liquidity ... 48

7.2.3 Investment opportunities ... 50

8. Conclusion ... 51

9. Appendices ... 54

9.1 Appendix I: Worldwide industrial copper usage ... 54

9.2 Appendix II: Worldwide industrial aluminum usage ... 54

9.3 Appendix III: Total imports of refined oil products (thousand barrels per day) ... 54

9.4 Appendix IV: Correlation among petroleum products ... 55

9.5 Appendix V: Commodity imports as percentage of total imports ... 55

9.6 Appendix VI: The 45 top chaebol of South Korea in 1997 ... 56

9.7 Appendix VII: Robustness test second-step regression ... 56

(5)

1 Introduction

1.1 Problem discussion

Firms engage in corporate risk management to increase firm value if there are capital market imperfections such as underinvestment problems (Gay and Nam, 1998; Froot, Scharfstein, and Stein, 1993), bankruptcy costs (Smith and Stulz, 1985), tax function convexities (Graham and Rogers, 2002; Smith and Stulz, 1985), or managerial risk aversion (Stulz, 1996; Stulz, 1984) present. Much of the empirical research on risk management has been concentrated on the relation between corporate hedging and firm characteristics, and tries to determine whether the behavior of firms that hedge is consistent with optimal hedging theories. He and Ng (1998) extend the risk management literature by comparing the effect of industrial

grouping membership on the hedging policies of firms. Based on their model, this paper helps shed light on the hedging literature by testing whether the effect of industrial group

membership on hedging incentives of firms is evident in a different institutional context. I also examine whether member firms‟ hedging incentives adjust in response to exogenous events which drastically change the institutional context. Linking the hedging incentives to commodity price exposure instead of the widely used exchange rate exposure provides new empirical insight in corporate risk management. Commodity prices are relevant for a

corporate risk management since these are more volatile than exchange rates and interest rates (Bartram, 2005) and hence can increase the economic vulnerability of firms to unforeseen shocks (Combes and Guillaumont, 2002).

1.2 Background

I analyze this effect for the 1994-2001 period with South Korean non-financial firms as the subject of my analysis because of the following three reasons. First, South Korea‟s economy largely depends on international trade. It had the highest trade dependence1 among the Group of 20 (G-20) leading countries during the 1980s and 1990s (World Bank, 2010). The export-dependency is due to the fact that the South Korean government desired an export-oriented approach to develop a competitive advantage for labor-intensive manufactured products on the world market. Yet, these products require the use of raw materials, such as crude oil and non-ferrous metals, which South Korean firms for a large extent have to import since South Korea lacks sufficient supply of natural resources. Next, South Korea‟s economic landscape is dominated by large industrial groupings, so-called chaebol. To see this, in 1995 chaebol accounted for approximately 16 percent of the Korean GNP, and this portion would be much

(6)

bigger if it included all the chaebol affiliates (Chang, 2003). For the same year, the average size of chaebol is about 10 times bigger than that of non-chaebol firms (Shin and Park, 1999). Finally, South Korea‟s institutional context changed drastically after the 1997 Asian financial crisis. Prior to the crisis chaebol did not practice any risk management since they assumed that the banks were always ready to roll over their debts or the government would step in to avoid bankruptcy of a large chaebol (Lee, 2000). The crisis, however, triggered corporate reforms and in January 1998 the government initiated the „Five plus Three Principles‟

regulation aimed to drastically reform the business practices of the dominant chaebol. Stricter regulations for borrowing and changes in ownership structure forced the chaebol to become more concerned about their financial situation. Accordingly, the following research questions are formulated:

Research question 1: Are the hedging incentives for chaebol-affiliated firms different compared to the hedging incentives for non-chaebol firms?

Research question 2: Do the hedging incentives for chaebol-affiliated firms change following the 1997 Asian financial crisis?

1.3 Method and data

(7)

2 Literature review

This section is divided into four parts. The first part reviews the theoretical explanations of why firms hedge and related empirical evidence. The second part describes the institutional context of South Korea before the 1998 corporate reforms and highlights three distinct characteristics of the South Korean business environment and their influence on the risk management behavior of South Korean non-financial firms, particularly chaebol. The third part focuses on the risk management practices after the 1998 corporate reforms of chaebol in particular. The last part, since the hedging incentives of South Korean firms is examined in relation to commodity price exposure, exemplifies the importance of commodities for the South Korean market.

2.1 Theoretical explanations of why firms hedge

A perfect capital market is characterized by (i) neutral taxes; (ii) no capital market frictions (i.e., no transaction costs, asset trade restrictions or bankruptcy costs); (iii) symmetric access to credit markets (i.e., firms and investors can borrow or lend at the same rate); and (iv) firm financial policy reveals no information. The Modigliani and Miller Theorem (1958) shows that corporate financing policy is irrelevant in the absence of market imperfections. It implies that investors can replicate a firm‟s financial actions without costs thus there is no incentive for a firm to hedge since shareholders can change their holdings of risky assets to offset any change in the firm‟s hedging policy (Smith and Stulz, 1985). However, when one of these assumptions fails, an imperfect capital market exists and if firms are exposed to economic risks in an imperfect environment these exposures impose costs on the firm (Graham and Rogers, 2002). For example, exposure to volatile commodity prices becomes costly. Firms can reduce these costs through hedging. The remainder of this paragraph identifies the economic incentives for a firm to hedge.

2.1.1 Reduction in expected taxes

There are two tax incentives for firms to hedge. First, Smith and Stulz (1985), and Mayers and Smith (1982) demonstrate that volatility is costly for firms with convex effective tax functions. As illustration, assume a firm is equally likely to lose $500,000 or earn $500,000 and that profits are taxed at 45 percent. Without hedging, even though expected income is zero, the firm expects to pay $112,500 in taxes2. Thus, tax function convexity provides a tax incentive to hedge if hedging reduces the variability of pre-tax firm values, subsequently the expected corporate tax liability is reduced and hence the expected post-tax value of the firm is

(8)

increased (Smith and Stulz, 1985).3 However, the empirical evidence regarding the tax function convexity is mixed. Mian (1996) finds weak evidence that hedging decisions are motivated by income tax saving strategies. Graham (2003) presents a similar result and interpret that firms do not hedge in response to convexity because the incentive is small relative to other hedging incentives. Second, by reducing the volatility of income, hedging increases debt capacity (e.g. Leland, 1998; Stulz, 1996). In response to greater debt capacity, firms add leverage and hence reduce tax liabilities (and increase firm value) due to an increase in interest deductions. Graham and Rogers (2002) find evidence that firms hedge to increase debt capacity and interest deductions which add approximately 1.1 percent to firm value. 2.1.2 Reduction in expected costs of financial distress

Mayers and Smith (1982), and Smith and Stulz (1985) suggest that hedging reduces the probability of incurring financial distress costs by reducing cash flow variability and thereby reducing the variance of firm value. The expected payoffs to the firm‟s claimholders are higher if the expected bankruptcy costs are reduced. To see this, I use the example of Smith and Stulz (1985). If the value of the firm is below the face value of debt, henceforth named F, at maturity, the bondholders receive F minus the transaction costs of bankruptcy. Otherwise, the shareholders receive firm value minus both taxes paid and the bondholders‟ payment, F. A risk management program that effectively eliminates expected bankruptcy costs reduces the variability of the future value of the firm and benefits shareholders. Thus, transaction costs of bankruptcy can induce widely held corporations to hedge.

Existing studies use leverage to proxy for the probability of financial distress, but results are mixed. Haushalter (2000), and Gay and Nam (1998), among others, confirm the hypothesis and find a positive relation between leverage and hedging whereas Nance et al. (1993) and Geczy, Minton, and Schrand (1997) do not find supportive evidence. By taking into account industrial grouping membership, He and Ng (1998) show that affiliated firms have a lower probability of financial distress and therefore have less incentive to hedge.

2.1.3 Avoidance underinvestment problem

The underinvestment problem arises during times when firms have to forego positive NPV projects, because external financing is too expensive and there are no sufficient internally generated funds to finance growth opportunities (Gay and Nam, 1998; Myers, 1977). Froot et al. (1993) indicate that the supply of internal funds can be disrupted by external factors such

3

(9)

as movements in commodity prices or exchange rates and that hedging can be used to create a more stable supply of internal funds. Empirical studies provide mixed support for the

underinvestment hypothesis. Geczy, Minton and Schrand (1995), and Dolde (1995)

demonstrate that firms with high R&D expenses are more likely to use derivative instruments whereas Mian (1996) finds a negative relation between a firm‟s investment opportunities and its derivative use.4 In an extensive study regarding the underinvestment problem, Gay and Nam (1998) use five different proxies of growth opportunities and their findings imply that firms‟ use of derivatives may partly be driven by the need to avoid potential underinvestment problems.

2.1.4 Managerial compensation, risk aversion, and hedging

Smith and Stulz (1985) illustrate that non-investor groups such as employees, managers, customers, and suppliers are typically unable to diversify away large financial exposures. If there is an increased probability of financial distress, these risk averse individuals require extra compensation to bear the non-diversifiable risk of the claims. For example, suppliers will be more reluctant to enter into long-term contract while employees demand higher wages if the probability of layoff is greater. To the extent risk management can protect the stakes of each of these corporate stakeholders, the firm can improve the terms on which it contracts with them and hence increase firm value (Stulz, 1996). The latter also claims that it is more difficult for owner (-managers) of private or closely-held companies to diversify away financial risks than shareholders of large public companies. Hedging financial exposures could add value by reducing the owners‟ risks.

2.2 Institutional context South Korea before corporate reforms

The subsequent paragraphs describe three distinct characteristics of the South Korean business environment before the 1998 corporate reforms. These characteristics are related to the hedging incentives of South Korean non-financial firms, particularly chaebol.

2.2.1 Corporate governance

The chaebol can be regarded as a controlling minority structure (CMS) in which it differs from other Asian business groups in that they combine effective family control with relatively low concentration of ownership (Haggard, Lim and Kim, 2003; Lee, K., 2000; La Porta, Lopez-de-Silanes, and Shleifer, 1999). Their subsidiaries are governed by a parent company, which in turn is owned by one family. The founding family holds the decision making right

4

(10)

through interlocking ownership among subsidiaries even though the concentration of in-group ownership decreased significantly after 1983 (see table 1). In turn, these „quasi holding

companies‟ indirectly or directly control most other member companies (Kim, 2003). Almeida, Park, Subrahmanyam and Wolfenzon (2010) demonstrate that chaebol use a pyramid structure in which a few central firms hold stakes in a large number of firms which protects against hostile takeovers and serves as an entry barrier.

Table 1.

In-group ownership share for the top chaebol (%)

Chaebol 1983 1987 1990 1995 2000

Top 30 57.2 56.2 45.4 43.3 43.4 Family 17.2 15.8 13.7 10.5 4.5 Subsidiaries 40.0 40.4 31.7 32.8 39.9 Top 5 n.a. 60.3 49.6 n.a. n.a. Family n.a. 15.6 13.3 n.a. n.a. Subsidiaries n.a. 44.7 36.3 n.a. n.a. Hyundai 81.4 79.9 60.2 60.4 n.a. Samsung 59.5 56.5 51.4 49.3 n.a. Daewoo 70.6 56.2 49.1 41.4 n.a. LG 30.2 41.5 35.2 39.7 n.a.

Note: The in-group ownership share for a chaebol is calculated by obtaining the weighted average of the combined ownership share for the founder‟s extended family and subsidiaries for all subsidiaries;

Source: Yoo (1999)

(11)

Existing studies link the likelihood of effective risk management to the strength of a firm‟s corporate governance. Allayannis, Lel, and Miller (2009) indicate that risk management is valuable only if it is coupled with strong governance. Lel (2009) finds that firms with strong corporate governance use currency derivatives for value-maximizing reasons whereas firms with weak corporate governance use the same derivatives mostly for managerial self-interests and selective hedging. A study of listed New Zealand companies illustrates that internal governance mechanisms, such as the board composition, together with the regulatory

environment also play a role in corporate derivative policy (Marsden and Prevost, 2005). With respect to chaebol, their board monitoring was ineffective as most directors were internal and represented interests of chairman and his family (Cho, 2004).

2.2.2 Government-business risk partnership

The chaebol dominate South Korean‟s economic landscape since the Park government launched the first five-year economic development plan (1962-1966) which also resembled the start of a government-business partnership. The government-led economic development strategy focused on expansion into targeted strategic industries and firms that would enter these selected industries received government support such as financial subsidies, tax reductions and debt guarantees.

Reasonably, firms diversified their operations in line with government industrial targeting policies (Jwa and Lee, 2004). This diversification process was mainly financed with

(12)

created a risk for the South Korean government. The cost of a possible failure of a chaebol grew and, from then on, the government relied on the „too big to fail‟ principle (TBTF).

The TBTF logic made chaebol take excessive risk, because they knew that the government would intervene when financial distress occurred. The government wanted to overcome a chaebol bankruptcy, because it could cause large-scale „ripple effects‟ such as major unemployment or liquidation of all related firms (Chang, 2000; Yoo, 1997). It created a situation where the chaebol were never really pressured by market discipline to restructure themselves resulting in moral hazard behavior by chaebol managers which directed them to accumulate vast amounts of debt to fund their investments (Lee and Rhee, 2007; Lee, K., 2000; Lee, J., 1998). He and Ng (1998) find that a similar situation existed in Japan during the 1990s where the major bank of the business group, keiretsu, prevents bankruptcy of a

financially distressed member firm.

Alternatively, other studies argue that the rescue of large enterprises by the Korean

government (e.g. Kia‟s bailout in 1997) should not be seen as bailouts in the strictest sense (Chang and Park, 2004; Chang, 2000). This is exemplified by the fact that in the beginning of the 1990s, three of the 30 largest chaebol went bankrupt (Chang, Park, and Yoo, 1998). They argue that the financial injections by the South Korean government were conditional on the change of ownership and top management of chaebol, and were always accompanied by tough terms of financial restructuring.

Financial liberalization in the 1980s changed the government-business risk partnership in the sense that the chaebol gained more independence in their major investment decisions. The government emphasized non-chaebol industries, such as mining and fishery, and encouraged loans to SMEs. Moreover, chaebol were confronted with stricter credit controls and bank privatization.

2.2.3 Bank relations and NBFIs

In order to promote the targeted industries during the period from the 1960s till mid-1980s, the government applied the lending criteria and also appointed the major bank‟s top

(13)

K., 2000; Lee, J., 1998). This government influenced credit allocation scheme resulted in over-capacity in several targeted industries, such as the HCI, and low returns on equity and high leverage for South Korean firms (Hahm, 2003; Joh and Ryoo, 2000). Empirical evidence also suggests that credit allocation to ailing firms does not improve long-term performance, but essentially increases the probability that the firm faces a loss (Lee, Peng, and Lee, 2004).

Banks neglected their responsibility to adequately examine the creditworthiness albeit

chaebol relied heavily on debt. As Campbell II and Keys (2001) point out, Asian banks made corporate loans with the implicit understanding the government would bail out the banks in the event of default while Johnson et al. (1999) claim a similar argument that South Korean banks may not have been effective in monitoring during the pre-crisis period. Likewise, banks had no incentive to monitor chaebol since their loans were backed by collateral and/or cross-guarantees by the member firms (Lee, K., 2000).

The financial liberalization in the 1980s limited the availability of bank credits for chaebol and forced them to find an alternative source of finance. Unlike the keiretsu in Japan, chaebol were not allowed to incorporate privatized commercial banks into their portfolio. The chaebol therefore turned to non-financial banking institutions (NBFIs) as an important financing source given that there existed no out-right ownership regulation for NBFIs. As a result, borrowings from NBFIs increased substantially during the 1980s and the share exceeded borrowings from the banking sector by early 1990s (Hahm, 2003). Through their NBFIs, the chaebol could also obtain credit from foreign source with foreign lenders not having

incentives to monitor Korean institutions, because they assumed to be protected from loss by the Korean government.

The chaebol-affiliated NBFIs also extended credit without carefully monitoring the financial risks. The government implied lenient regulations to NBFIs, compared to the strictly

(14)

systematically borrowed more at lower borrowing costs than chaebol that did not own an NBFI. Additionally, the TBTF logic influenced the behavior of NBFIs. They regarded the financial risk of chaebol, despite decreasing profitability, as not so high due to continuous government intervention in credit markets. The top 30 chaebol were, as a consequence, able to increase their debt-equity ratio to 519.0 percent in 1997 (Kwon and Shepherd, 2001).

2.3 Institutional context after corporate reforms

Different South Korean governments have tried to improve the internal and external governance of the chaebol. Throughout the years different ineffective measures were undertaken that tried to remove the agency problems resulting from the chaebol ownership structure. For example, the establishment of the Monopoly Regulation and Fair Trade Act (MRFTA) in 1980 had the goal to limit the concentration of economic power, but failed to maintain institutional reforms. Serious reforms of the chaebol governance structure finally started in the wake of the 1997 financial crisis.

The Kim Dae-jung government had been assigned with the difficult task to start corporate reforms after South Korea was seriously hit by the 1997 financial crisis. On January 13, 1998 President Kim and chaebol leaders agreed on five principles5 of corporate restructuring. On August 15, 1999 President Kim announced three supplementary principles6. These „Five plus Three‟ principles aimed to destroy traditional characteristics of the chaebol and to build an Anglo-American corporate governance system (Mo and Moon, 2003; Yanagimachi, 2004). The chaebol, once regarded as drivers of South Korea‟s export-oriented economic growth, were now blamed for poor corporate governance, high leveraging and irresponsible

diversification.

One of the most important measures was to discourage corporate over-borrowing. The five largest chaebol were asked, and succeeded, to reduce their debt ratios below 200 percent by the end of 1999 and to dismantle their debt guarantee practices (Lee, JW., 2004).

Accumulation of credit through their NBFIs also became more difficult, because these institutions faced more stringent monitoring by independent directors. Borensztein and Lee

5

The five principles were: (i) hold chaebol leaders more accountable for managerial performances (ii) boost managerial transparency (iii) improve financial health (iv) focus on core businesses (v) eliminate loan guarantees among affiliates.

6

(15)

(2002) demonstrate that chaebol-affiliated firms appear to have lost the preferential access to credit that they had enjoyed in the pre-crisis period. Moreover, the corporate reforms also resolved the mutual debt guarantees among chaebol affiliates.

Williamson (1975) argues that the internal market of the conglomerate firm can improve capital allocation and more effective management of divisions if external markets are non-existent or poorly performing. The chaebol-affiliated firms also benefited from conglomerate membership and Chang and Hong (2000), and Shin and Park (1999) find evidence that there exists an internal capital market for chaebol which reduces their financing constraints7. Zeile (1996) demonstrates that chaebol-affiliated firms are more dominant in the industries that are subject to market failures than others. However, the prohibition on mutual debt guarantees, circular investment, and unfair transactions among chaebol affiliates made it difficult for chaebol to transfer and share financial resources and management know-how across subsidiaries (Kim, Hoskisson, Tihanyi, and Hong, 2004).

The corporate reforms also remarked the end of the TBTF mentality. The South Korean government forced chaebol into bankruptcy (e.g. Daewoo in 1999). The lost of full government support leads, according to Hong et al. (2004), to a rise in the number of acquisitions to avoid bankruptcy. The corporate governance system of chaebol underwent major changes that led to more effective monitoring of management. Monitoring mechanisms improved through strengthened minority shareholders rights, more independency for the board of directors, and increased voting rights for institutional investors (Cho, 2004). The stake of foreign shareholders in chaebol increased substantially and this fact could not be disregarded and required more transparency for chaebol in their management and accounting practices (Lee, JW., 2004).

2.4 Relevance commodity prices for South Korean firms

For decades, South Korea is an important player on the world market. The export-oriented economic growth during the decades before the 1997 Asian crisis made South Korea the twelfth largest importer as well as exporter in world merchandise trade (WTO, annual report 1998). The high ranking of South Korea in terms of imports is mainly due to South Korea‟s poor resource endowment. South Korean firms therefore have to import large quantities of commodities including crude oil, mineral fuels and non-ferrous metals (Bureau of East Asian

1

(16)

and Pacific Affairs, 2008). The high ranking in terms of exports is attributed to South Korea‟s dependency on exports to fuel the economy. The major exports include machinery, transport equipment, and electrical equipment. The dependence of South Korean‟s economy on imports and exports makes the country vulnerable for commodity price volatility.

On a national level, the economic vulnerability can have a negative impact on South Korea‟s economic growth, its development or welfare (Combes and Guillaumont, 2002). For example, petroleum accounts for about 18 percent of South Korea‟s total imports in 2000 (see figure 1) and the volatility of the petroleum price can therefore be a major source of instability and uncertainty (Larson, Varangis, and Yabuki, 1998). Additionally, commodity prices can also have a direct effect on South Korean‟s terms of trade or GDP growth (Abeysinghe, 2001; Backus and Crucini, 2000).

Figure 1.

Imports of petroleum, natural gas, iron and steel and non-ferrous metals as percentage of total imports

Source: Korean Statistical Information Services

On a firm level, commodity price volatility can have an impact on firm value. For example, oil price changes appear to affect a firm‟s stock returns (Nandha and Hammoudeh, 2007; Basher and Sadorky, 2006). Existing studies have shown that firms aware of their commodity price exposure may engage in hedging activities (Haushalter, 2000; Tufano, 1998; Edwards and Canter, 1995). In a survey about the hedging practices of South Korean firms, Pramborg (2005) finds that South Korean firms, though to a lesser extent than foreign currency

derivatives and interest rate derivatives, make use of commodity derivatives. Alternatively, in case commodity prices affect the costs and revenues of firms only to a small degree,

(17)

commodity price changes can be passed on to customers or other firms, so-called price pass-through (Bartram, 2005).

Industries should be differently affected by commodity price risk since the underlying asset can be an input or output factor in the production process of the firm. For instance, copper is an important input for the manufacturing process for firms in the electrical equipment industry, but is of less importance for firms in the oil and gas industry. Similar to Bartram (2005), it is assumed that commodity price exposures can be measured empirically in a particular industry according to its major input/output relationships of commodities. Reasonably, an input factor should result in a negative commodity price exposure, while exploited as an output factor should lead to a positive exposure (see table 2).

Table 2.

Hypotheses on commodity price exposures by industry

Agriculture/forestry wheat (+/–), barley (+/–), oats (+/–), cattle (+), hogs (+) Public utilities/mining crude oil (–), natural oil (–)

Chemicals gasoline (+/–), heating oil (+/–), diesel (+/–) Rubber/plastics rubber (+), crude oil (–)

Primary metal aluminum (–), copper (–), zinc (–), lead (–), nickel (–),\ tin (–) Industrial machinery aluminum (–), copper (–), titanium (–), rubber (–)

Transp. equipment aluminum (–), copper (–), transportation equipment (–), titanium (–) Elect. equipment lead (–), copper (–), mercury (–), silicon (–), selenium (–), tungsten (–) Misc. manufacturing aluminum (–), plate (–), copper (–), zinc (–)

Paper/publishing paper pulp (–), paper (+/–) Textile/leather cotton (+/–), jute (+/–)

Food/tobacco barley (–), wheat (–), coffee (–), sugar (–)

Note: The table reports exposure hypotheses for different industry sectors based on major input/output relationships of various commodities. The signs in brackets indicate the expected direction of the exposure (+: positive, +/–: unclear direction, –: negative).

Source: Bartram (2005)

2.5 Conclusion

South Korean firms import large quantities of commodities, because South Korea lacks

sufficient supply of raw materials. The demand for commodities increased from the 1960s and onwards when the government forced firms to adopt an export-oriented approach for

manufactured products. This should eventually lead to a competitive advantage in these products on the world market. In pursuing this goal, the government extended subsidies and cheap credit to South Korean firms, except these policies were largely skewed in favor of the chaebol. The chaebol could therefore grow as large that their bankruptcy would have

(18)

continued with borrowing large sums of debt. The profitability of the chaebol deteriorated over time, but the weak South Korean legislative system and weak chaebol corporate governance system created no incentives for chaebol owner-managers to care more about profit maximization and less about their own private interests. For these reasons, I claim that chaebol owner-managers care less about their financial situation than non-chaebol managers and hence have less incentive to hedge.

Hypothesis 1: Chaebol-affiliated firms have less incentive to hedge than non-chaebol firms.

Stricter regulations for borrowing and changes in ownership structure forced the chaebol to become more concerned about their financial situation after the 1997 Asian financial crisis. The adoption of a more Anglo-American corporate governance structure stimulated chaebol to pursue interests other than private benefits. Hypothesis 2 is therefore as follows:

(19)

3. Methodology

3.1 Model description

I apply the two-step regression model as used by He and Ng (1998)8. The first-step regression is a two-factor OLS and determines the magnitude of the different commodity price exposures. This first-step regression test is widely used in previous commodity exposure studies, thus it allows for comparability of results (e.g. Bartram, 2005; Faff and Brailsford, 1999; Tufano, 1998).

(1)

In this model, represents the weekly stock return9 of firm i in period t, designates the

overall return on the KOSPI capital market index m in period t, signifies the unexpected weekly change in the commodity price or commodity index for the period t and denotes

the white noise error term. Hence, is interpreted as the commodity price exposure coefficient. The estimation is net of hedging, because describes the sensitivity of stock returns to unforeseen changes in commodity prices.

The definition of commodity price exposure refers to the unexpected change in the

commodity price or commodity index due to the fact that price variability of commodities makes it difficult forecasting future commodity prices as well as future commodity prices are subject to large and unpredictable movements which may have persistent effects (Cashin and McDermott, 2002). Pindyck and Rotenberg (1990) find that macroeconomic shocks influence the movement of commodity prices, but indicate that these macroeconomic shocks occur unexpectedly. French (1986) has a similar argument and argue that the true expectation of the future spot price is unobservable. For these reasons, the volatility of commodity spot prices can be regarded as predominantly unanticipated by firms.

The above-mentioned model Eq. (1) presumes the presence of constant variance (i.e.

homoskedasticity). Yet, it is unlikely in the context of financial (weekly) time series that the variance of the errors will be constant over time (Brooks, 2008). Engle (2001) describes this as data in which the variances of the error terms are not equal, so-called heteroskedasticity (or

8

The second-step regression slightly differs from He and Ng‟s approach. I therefore conduct a robustness check using their exact model to validate my results (see Section VII).

9_{Logarithmic rather than arithmetic returns are used because their distribution is found to be more nearly normal.}

(20)

an ARCH model, which stands for autoregressive conditionally heteroskedastic). The standard error estimates could be wrong if the error terms are assumed homoskedastic when they in fact are heteroskedastic. For instance, in their study about Asian foreign exchange exposure, Muller and Verschoor (2007) find that from more than 96 percent of the firms in their sample the error terms are heteroskedastic. Therefore it makes sense to consider a model that does not assume a constant variance. As suggested by Brooks (2008), and as applied by Muller and Verschoor (2007), a GARCH (1,1) specification is added to the first-step regression model Eq. (1)10. It is more appropriate to use a GARCH (1,1) specification to correct for heteroskedasticity instead of the ARCH model because the former is more

parsimonious, and avoids over fitting (Brooks, 2008). Accordingly, I use the following regression model with the GARCH (1,1) specification for the first step:

(2)

with _and

(3)

where depicts the conditional variance of the residuals and the white noise error term.

In Eq. (2), the term serves the purpose of a control variable for all other systematic effects impacting stock price. According to Jorion (1990), the influence of macroeconomic variables on stock returns should be included in the exposure model to overcome exaggerated estimates attributable to foreign currency movements. Chue and Cook (2008) regard the market index as a way to „soak up‟ any remaining correlation between other macroeconomic factors and .

The second-step cross-sectional regression tests the effect of the determinants that control for hedging on the commodity price exposure of South Korean firms11. The cross-sectional regression specification makes use of the firm-level weekly commodity price exposure beta as the dependent variable estimated over the 1994-1997 and 1998-2001 sub-periods. The

explanatory variables consist of firm-level information.

(4)

10_{I used the Engle (1982) test for ARCH effects to make sure that the inclusion of a GARCH (1,1) specification}

is appropriate. I found that 89 percent of the firms in my sample the error terms are heteroscedastic.

11_{I use SPSS for the second-step cross sectional regression, because the regression results display the}

standardized coefficients. Standardization enables me to compare directly the relative effect of each independent variable on the dependent variable (Hair et al., 2009).

(21)

is the value of the commodity price exposure of firm i, and represents long-term debt ratio, denotes the dividend payout ratio, signifies the quick ratio, represents the ratio of capital expenditures to total assets, and represents the natural logarithm of total assets estimated for firm i.

Chamberlain, Howe, and Hopper, (1997) show that moving from monthly to daily stock return data makes it easier to discern firm-level exchange rate exposure. However, the use of daily frequency data can suffer the bias of infrequently traded stocks (Tufano, 1996) whereas the use of monthly data will result in too few observations. Particularly for the four year sub-periods, I would encounter more difficulties to find firm level commodity price exposure. In order to overcome the aforementioned potential problems, and taking into account the

foundation of my research, stock return data is obtained on a weekly basis as this provides the most accurate results.

3.2 Operationalization variables

3.2.1 First-step regression

Table 3 presents the operationalization of the variables for the first-step regression. Table 3.

Operationalization variables first-step regression

Variables _Measurement

1 Dependent Company stock return Total weekly return on a firm's stock 2a

Independent

Unexpected changes of the crude oil price Weekly return from holding 1 barrel of Dated Brent based on New York Mercantile spot price

2b Unexpected changes of the aluminum price Weekly return from holding 1 metric ton aluminum based on London Metal Exchange spot price

2c Unexpected changes of the copper price Weekly return from holding 1 ton copper based on London Metal Exchange spot price

2d Unexpected changes of the GSCI Energy index Weekly return on S&P Energy index

2e Unexpected changes of the GSCI Industrial Metals index Weekly return on S&P Industrial Metals index 2f Unexpected changes of the GSCI Agricultural index Weekly return on S&P Agricultural index

3 Control Market stock return12 Total weekly return on KOSPI index

Many South Korean firms use commodities as an input for the manufacturing process. Hence, commodity price fluctuations are an important source of risk for South Korean firms as this

12

(22)

can lead to fluctuations in cash flows and accounting earnings. It therefore appears warranted to analyze the sensitivity of South Korean firms‟ stock prices vis-à-vis commodities prices. The commodity price exposure is examined using copper, aluminum, and crude oil prices. The individual copper exposure of South Korean firms is analyzed because copper is used in many different industries for a wide array of products. Copper is an important base metal for firms in the electrical and building industries since copper is an excellent conductor of electricity. It is also used in large quantities by the construction industry (see Appendix I).

The individual aluminum exposure of South Korean firms is analyzed, because it is the most extensively used metal in the world. Since aluminum is extraordinarily strong, it is

fundamental to both the automobile and the air travel industries. Furthermore, aluminum key features are also used for packaging and construction due to the fact that it does not rust (see Appendix II).

Crude oil‟s (i.e. petroleum) biggest value is its wide use as a raw material in the chemical industry. In addition, petroleum is important for the generation of electricity and

manufacturing of a broad selection of products. South Korean firms use large quantities of refined petroleum as is highlighted by South Korea‟s top 10 rank as top world net importer of oil throughout the 1990s and 2000s (see Appendix III).

Commodity price indices are included, because prices of commodities from the same category are most probably correlated. For example, Appendix IV shows the correlation among

petroleum products. The GSCI S&P commodity indices aggregate different commodities of a similar type into an index and are also used by Bartram‟s study (2005) about commodity price risk13. I include the energy index (crude oil, Brent crude oil, unleaded gas, heating oil, gasoil, natural gas), metals price index (copper, aluminum, nickel, zinc, lead) and agricultural index (wheat, red wheat corn, soybeans, cotton, sugar, coffee, cacao).

3.2.2 Second-step regression

Table 4 gives an overview of the variables for the second-step regression. I conduct the second-step regression analysis using a weighted commodity price index consisting of the beta regression coefficients of aluminum, copper and crude oil, because this better reflects the significance of the commodities as an input/output factor for South Korean firms. The weight

13

(23)

Table 4.

Operationalization second-step variables

Variables Measurement

1

Dependent

Weighted commodity price exposure index

Beta regression coefficient of the effect of changes in the aluminum prices on the firm stock returns Weight: 0.86

Beta regression coefficient of the effect of changes in the copper prices on the firm stock returns Weight: 0.057

Beta regression coefficient of the effect of changes

in the crude oil prices on the firm stock returns

Weight: 0.083

2a

Independent

Firm leverage

Four-year average of the ratio of year-end book value of long-term debt to the sum of the book value of equity and book value of debt

2b Firm liquidity position Four-year average of the dividend payout per share divided by _{earnings per share of the firm in Korean Won}

2c Firm liquidity position Four-year average of the Quick Ratio of the firm

2d Investment opportunities Four-year average of the ratio of capital expenditures to

book value of total assets

2e Firm Size Natural logarithm of the four-year average of the total assets of _{the firm in Korean Won}

of crude oil, copper, aluminum is based on the classification of the S&P GSCI in which the weight of each commodity is determined by the average quantity of production as per the last five years. The S&P GSCI weights are 0.86, 0.057, and 0.083 respectively and reflect the relative significance of each commodity in the world economy. I also considered the relative importance of each commodity for the South Korean economy by looking at the imports as well as exports of crude oil and non-ferrous metals as percentage of total imports and total exports. Crude oil imports are most important for the South Korean economy (see Appendix V). For this reason, I apply the weights as prescribed by the S&P GSCI.

I use leverage to proxy for the probability of the firm encountering financial distress which is directly related to the size of the firm‟s fixed claims relative to its assets (Nance et al., 1993). Financial distress is seen as costly, because it creates a tendency for firms to do things that are harmful to debt-holders and non-financial stakeholders (i.e. customers, suppliers, and

(24)

compute leverage as the firm‟s long-term debt ratio. A comparable yearly measure of a firm‟s debt-size ratio is applied by He and Ng (1998), Tufano (1998), and Mian (1996).14

Firms could reduce the possibility of encountering financial distress by maintaining a larger short-term liquidity position through investment in more liquid assets or a lower dividend pay-out ratio which helps to assure bondholders that funds will be available to pay fixed claims (Nance et al., 1993). Therefore, firms with a high short-term liquidity position are less likely to engage in hedging activities since they have less financial constraints (Nguyen and Faff, 2002; Mello and Parsons, 2000). Two different variables are used as proxies for a firm‟s term liquidity position. First, the quick ratio measures a firm‟s ability to meet its short-term obligations with its most liquid assets such as cash (Geczy et al. 1997). The quick ratio is regarded a better determinant for short-term liquidity than the current ratio, because it more accurately captures the internal wealth of a firm (Berman and Bradbury, 1996). Second, the dividend payout per share might also influence a firm‟s hedging policy. Froot et al. (1993) argue that high-dividend payers are not likely to be liquidity constrained since they have chosen to pay out cash rather than use it for investment.

Recall from Section II that another theoretical explanation of why firms hedge concerns the alleviation of the underinvestment problem. It concerns that external financing is sufficiently expensive that firms must reduce investment spending during times when internally generated cash flows are not sufficient to finance growth opportunities (Gay and Nam, 1998). Hence, firms are more likely to hedge by the need to avoid potential underinvestment problems. I use a ratio of capital expenditures over total assets as a proxy for a firm‟s future investment opportunities (Carter, Rogers, and Simkins, 2003; Lang, Ofek, and Stulz, 1996). This measures net investment.15

Nance et al. (1993) suggests that firm‟s incentives to hedge are also affected by firm size, but that the effect is ambiguous and should empirically be determined. For instance, they argue that firm size is positively related to hedging incentives since larger firms have economies of scale in hedging costs. On the other hand, smaller firms could have more incentive to hedge,

14_{In accordance with other studies, including Haushalter (2000), I also examined the ratio of the book value of}

short-term and long-term debt to the market value of assets. The analysis result is quantitatively similar when I use this alternative proxy.

15_{Two other common used proxies to capture a firm‟s investment opportunities are market-to-book-value ratio}

(25)

(26)

4. Data description

4.1 Selection procedure

The selection procedure for my sample consists of four steps:

1. Identification of the firms that were listed on the Korean Stock Exchange as of January 1, 1994 as provided by Korean Listed Company Association (KLCA) and DataStream International. The latest available data results in a total of 423 firms. 2. 169 firms are excluded from the research sample for the following reasons:

a. Unavailability stock price data;

b. Unavailability financial statement data;

c. Merger with another firm in the period 1994-2001; d. Acquired by another firm in the period 1994-2001; e. Bankruptcy in the period 1994-2001.

3. Previous commodity exposure studies predominantly rely on the internationally accepted Standard Industry Classification (SIC) (e.g. Jin and Jorion, 2006; Bartram, 2005). However, this classification appears incomplete for several South Korean firms. The industry classification by the KLCA is therefore used and divides the sample firms in 10 industry classes. I exclude 62 firms from the financial, distribution, services and textile and wearing apparel industry since there is an unclear link with specific commodity dependency. The final sample consists of 107 firms in 6 different industries (see table 5).

4. In line with previous studies about chaebol, I use the yearly publication of the Korean Fair Trade Commission (KFTC) to identify the chaebol and their affiliates (e.g. Joh, 2003; Kim and Lee, 2003). The KFTC ranks the top 45 chaebol and their affiliates according to their combined value of total assets (see Appendix VI). Accordingly, my sample consists of 60 chaebol-affiliated firms and 47 non-chaebol firms (see table 5).16

16

(27)

Table 5.

Research sample: Chaebol and industry classification

Chaebol

Non-Chaebol

Machinery & Construction 11 7 18

Electrical and Electronic Equipment 11 9 20

Iron and Metal Products 7 11 18

Transport Equipment 12 4 16

Chemicals 14 10 24

Foods & Beverages 5 6 11

Total 60 47 107

4.2 Sampling period

Even though there is more recent data available for South Korean firms, I opt for a sampling period from 1994 till 2001. I want to examine whether the effect of industrial grouping membership adjusts over time due to changes in the institutional context. The 1998 structural reforms change the competitive landscape in South Korea in the sense that chaebol suddenly face financial and financing constraints due to increased government oversight, stricter credit regulations, and major changes in the corporate governance system. The chaebol are as a consequence more exposed to economic risks in an imperfect environment which could provide incentives for chaebol to engage in (enlarged) hedging activities.

4.3 Descriptive Statistics

4.3.1 Summary statistics commodities

The mean log price changes for aluminum, crude oil, GSCI energy, and GSCI industrial metals are positive which implies that these commodity prices increased in the 1994-2001 period (see table 6). This result is not surprising since crude oil has a large weight in the GSCI energy index and aluminum a large weight in the GSCI industrial metals index. On the other hand, the mean log price changes for copper and GSCI agriculture are negative and indicate that these commodity prices decreased in the 1994-2001 period. The small standard deviation values show that there is not much dispersion from the average. This notion, together with the skewness and kurtosis statistics (i.e. near zero), suggest that the distribution of the

(28)

Table 6.

Summary statistics for the continuously compounded weekly commodity price movements

Aluminum Copper Crude Oil GSCI Energy GSCI Industrial Metals GSCI Agriculture Mean 0.000198 -0.000232 0.000406 0.000386 0.000013 -0.000842 Median -0.001134 -0.000609 0.002207 0.000667 -0.000296 -0.001246 Min -0.034774 -0.059755 -0.077509 -0.074587 -0.028599 -0.035748 Max 0.035987 0.044714 0.065708 0.044173 0.027510 0.035810 Std Dev 0.010167 0.012389 0.022147 0.016983 0.009297 0.008945 Skewness 0.225574 -0.106782 -0.393532 -0.467679 0.122437 0.181404 Kurtosis 0.569569 1.128113 0.805965 1.570467 0.101163 1.131994 Observations 417 417 417 417 417 417

Note: Sample period is January 7, 1994 to December 28, 2001

4.3.2 Descriptive statistics South Korean firms

The total assets point out that the research sample consists of some large firms. On average, chaebol are significantly larger than non-chaebol firms (see table 7). The quick ratio indicates that, on average, South Korean firms have a low liquidity. Chaebol firms show a lower quick ratio than non-chaebol firms which can be due to the fact that chaebol have easy access to credit through their NBFIs and access to internal sources of financing (Baek, Kang, and Park, 2004). A similar situation exists for keiretsu firms in Japan where the group firms are partly owned by the bank within the industrial group. Since the bank finances all group firm‟s projects, keiretsu firms are less affected by a lack of liquidity than independent firms (Hoshi, Kashyap, and Scharfstein, 1991).

In the period prior to the crisis, the investment ratio of chaebol is significantly larger than those of non-chaebol firms (Shin and Park, 1999) whereas chaebol exhibit a sharp reduction in investment in the period after the crisis (Hong et al., 2004). Contrary to my expectation, I find that the size of the investment ratio, measured as capital expenditures over total assets, is comparable for chaebol and non-chaebol firms. A possible explanation could be that I present my descriptive statistics for the full sample period and make no distinction between the pre-crisis and post-pre-crisis period.

(29)

Again, I make no distinction between the pre-crisis and post-crisis period. Yet, results beyond the outcomes obtained in table 7 indicate that the debt-equity ratio of chaebol versus non-chaebol firms is significantly different in the pre-crisis period, but is not significantly different in the post-crisis period. The latter could be clarified with the fact that the chaebol, through the adopted debt restructuring programs, as well as SMEs, the government decided to roll over their debts, were able to lower their debt-equity ratio (Ahn, 2001).

Table 7.

Descriptive statistics South Korean Firms 1994-2001

All firms N Mean Lower

Quartile Median

Upper Quartile Total Assets 107 3,001,116 575,125 1,263,132 2,985,467 Capital Expenditures 107 259,613 30,818 81,427 181,098

Long-term Debt Ratio 107 2.21 1.47 1.82 2.42

Quick Ratio 107 0.73 0.51 0.62 0.83

Dividend Payout Ratio 107 20.68 9.13 20.11 30.05 Capital Expenditures / Total Assets 107 0.074 0.066 0.047 0.091

Chaebol N Mean Lower

Upper Quartile

Significantly different (t-values) Total Assets 60 3,592,078 758,377 2,051,593 3,827,451 Yes (2.1396)

Capital Expenditures 60 262,165 56,844 139,600 230,173 No (1.1839)

Long-term Debt Ratio 60 2.13 1.57 2 2.55 Yes (2.1762)

Quick Ratio 60 0.65 0.5 0.6 0.73 Yes (2.3528)

Dividend Payout Ratio 60 19.52 6.89 18.62 27.23 No (1.0247)

Capital Expenditures / Total Assets 60 0.077 0.045 0.065 0.087 No (0.5893) Non-chaebol N Mean Lower

Upper Quartile

Significantly different (t-values) Total Assets 47 2,186,903 373,370 855,240 1,263,133 Yes (2.1396)

Capital Expenditures 47 256,099 12,801 50,871 123,397 No (1.1839)

Long-term Debt Ratio 47 2.31 1.41 1.68 2 Yes (2.1762)

Quick Ratio 47 0.83 0.53 0.77 0.9 Yes (2.3528)

Dividend Payout Ratio 47 22.29 11.11 22.26 32.42 No (1.0247)

Capital Expenditures / Total Assets 47 0.07 0.049 0.066 0.096 No (0.5893) Note: Total assets and capital expenditures are in thousands of South Korean Won. Sample period is January 7,

1994 to December 28, 2001. The last column compares the means of chaebol and non-chaebol firms; t-values between parentheses.

4.4 Reliability and correctness model

(30)

and the market stock returns exhibit the highest correlation, 0.106, which is significant at the 0.05 level. The other commodity prices display no significant correlation coefficients. Table 8.

Correlation between the independent and control variables for the first-step regression

Alum inu m Co pp er Crude O il G SCI Ag ricult ure G SCI E nerg y G SCI I n d. M et a ls . Pearson coefficients 0.012 0.05 0.106* 0.025 0.042 0.029 Sig. (2-tailed) 0.81 0.309 0.031 0.608 0.395 0.559 Observations 417 417 417 417 417 417

* Correlation is significant at the 0.05 level (2-tailed)

Table 9 outlines the correlation matrix for the independent variables of the second-step regression. For chaebol, all significant correlation coefficients remain below the 0.500

threshold value. The quick ratio shows the highest significant positive correlation with capital expenditures in the 1994-1997 and 1998-2001 sub-periods (.443 and .447 respectively; significant at the 0.01 level). For non-chaebol, three correlation coefficients exceed the 0.500 threshold value. Size demonstrates significant negative correlation coefficients with the quick ratio in the 1994-1997 and 1998-2001 sub-periods (-.541 and -.519 respectively; significant at the 0.01 level). Size also shows a significant negative correlation coefficient with capital expenditures in the 1994-1997 sub-period (-.533; significant at the 0.01 level). Since the correlation coefficients that exceed the threshold value are close to .500, I conduct an additional multicollinearity analysis with the Tolerance and Variance Inflation Factor (VIF) values to verify whether it is still justified to retain all variables for the second step regression.

Table 10 presents the Tolerance level and VIF for my five independent variables. Tolerance is defined as the amount of variability of the selected independent variables not explained by the other independent variables. The high Tolerance value means a small degree of

multicollinearity. As a rule of thumb, if tolerance is less than 0.20, a problem with

(31)

though three Pearson correlation coefficients slightly exceed the 0.500 threshold value. Yet, I take the existing correlations into consideration when analyzing the results of the second-step regression.

Table 10.

Tolerance and VIF values for the independent variables of the second-step regression

Chaebol

1994-1997 1998-2001

Tolerance VIF Tolerance VIF

QR .740 1.351 QR .694 1.44 DEBT .850 1.177 DEBT .938 1.066 DIV .845 1.184 DIV .902 1.109 logSIZE .936 1.068 logSIZE .828 1.208 CAPEX .707 1.414 CAPEX .748 1.337 Non-chaebol 1994-1997 1998-2001

Tolerance VIF Tolerance VIF

QR .601 1.664 QR .707 1.414

DEBT .795 1.257 DEBT ,923 1.083

DIV .875 1.143 DIV .836 1.196

logSIZE .540 1.852 logSIZE .691 1.448

(32)

Table 9.

Correlation between the independent variables of the second-step regression

Chaebol

1994-1997 1998-2001

QR DEBT DIV LogSIZE CAPEX QR DEBT DIV LogSIZE CAPEX

QR 1.000 -.237 .230 -.041 .443** QR 1.000 -.136 .248 -.340** .474**

Sign. (2 tailed) .063 .063 .750 .000 Sign. (2 tailed) .291 .0.52 .007 .000

DEBT 1.000 -.229 .177 .003 DEBT 1.000 .047 .236 -.120

Sign. (2 tailed) .074 .169 .982 Sign. (2 tailed) .718 .065 .352

DIV 1.000 -0.092 .318* DIV 1.000 .020 .245

Sign. (2 tailed) .527 .012 Sign. (2 tailed) .879 .055

LogSIZE 1.000 0.160 LogSIZE 1.000 -.233

Sign. (2 tailed) .213 Sign. (2 tailed) .068

CAPEX 1.000 CAPEX 1.000

Non-chaebol

1994-1997 1998-2001

QR DEBT DIV LogSIZE CAPEX QR DEBT DIV LogSIZE CAPEX

QR 1.000 -.358* -.106 -.541** .268 QR 1.000 -.156 -.129 -.519** -.207 Sign. (2 tailed) .016 .487 .076 .000 Sign. (2 tailed) .307 .398 .000 .172

DEBT 1.000 -.217 .078 -.028 DEBT 1.000 -.196 .058 -.103

Sign. (2 tailed) .151 .856 .611 Sign. (2 tailed) .198 .703 .499

DIV 1.000 .195 .249 DIV 1.000 .177 .358*

Sign. (2 tailed) .100 .199 Sign. (2 tailed) .244 .016

LogSIZE 1.000 -.533** LogSIZE 1.000 .296*

Sign. (2 tailed) .000 Sign. (2 tailed) .045

CAPEX 1.000 CAPEX 1.000

(33)

5. Empirical results first-step regression

5.1 Commodity price exposure: 1994-2001

Table 11 presents the commodity price exposures for South Korean firms estimated with Eq. (2) and Eq. (3) for the full sample period. For the individual commodities, the first-step regression provides significant exposure coefficients at the 0.0517 significance level for the sample firms in the range 9.3% to 21.5%. Similarly, the commodity indices show significant exposure coefficients at the 0.05 level for the sample firms in the range of 8.4% to 12.1%. Table 11.

Commodity price exposure 1994-2001

(See note)

1994-2001

- + +/- Mean18 Median Minimum Maximum Aluminum 9 (8.4%) 6 (5.6%) 15 (14.0%) -0.0307 -0.0573 -1.2824 1.7261 107 Crude oil 13 (6.5%) 16( 15.0%) 29 (21.5%) 0.0391 0.0732 -0.6903 0.8198 107 Copper 6 (5.6%) 4 (3.7%) 10 (9.3%) -0.0446 -0.0357 -1.3530 0.8922 107 Agriculture index 6 (5.6%) 3 (2.8%) 9 (8.4%) 0.0112 0.0490 -1.5347 1.6913 107 Energy index 2 (1.9%) 10 (9.3%) 12 (11.2%) 0.0592 0.0138 -0.9389 0.7779 107 Industrial Metals index 8 (7.5%) 5 (4.7%) 13 (12.1%) -0.0334 -0.0407 -1.6076 1.9106 107

Note: The table reports the percentage of nonfinancial South Korean firms that show a significant commodity price exposure for different commodity price variables (5% level). The three columns refer to the number and

percentages of negative, positive and all exposures respectively. „ ‟ represent the total number of firms

included in the regression.

Note:

Even though South Korean firms still prefer the use of foreign exchange derivatives and interest derivatives over commodity derivatives (International Swaps and Derivatives Association, 2009) I find that for South Korean firms the foreign exchange risk is not of greater statistical importance than their commodity price risk. My results indicate that the commodity price risk for South Korean firms is in the range of 8.4%-21.5%, while Miller and Reuer (1998) show that only 14.6% of the South Korean firms are sensitive to exchange rate changes, and Muller and Verschoor (2007) conclude that 21.6% and 22.6% of 216 South Korean firms are exposed to the US dollar and Japanese yen respectively.19 Thus, it seems that, while South Korean firms import large quantities of commodities, commodity price

17

Stigler (1986) argues that choosing the level of significance is an arbitrary task but that with the use of a 5% significance level you have reasonable power to detect effect-sizes that are of interest. Moreover, a 5% significance level is generally used in existing hedging and exposure studies (e.g. Muller and Verschoor, 2007; Judge, 2006; Bartram, 2005).

18

Mean, Median, Minimum, and Maximum represent significant as well as non-significant betas.

19_{Another study by Lee (2003) found that about 20% of the Korean companies‟ stock performance is}

(34)

movements have less effect on their stock price than anticipated beforehand. There are several potential explanations for this result (i.e. small residual commodity price exposures). South Korean firms may engage in corporate hedging when commodities are an important input for the manufacturing process. In a survey about the hedging practices of South Korean firms (versus Swedish firms), Pramborg (2005) finds that 73 percent of the South Korean respondents makes use of derivatives. South Korean firms may also be naturally hedged. Judge (2003) illustrates such a situation with the following example about Rio Tinto: „Rio Tinto‟s exposure to commodity prices is naturally diversified by virtue of its broad

commodity spread, and the Group does not believe a commodity price hedging program would provide long term benefit to shareholders.‟ Furthermore, commodity prices may affect only few corporate cash flows, resulting in a small effect relative to firm size (Bartram, 2005). The movements of the individual commodity prices have a different impact on the stock returns of South Korean firms. For instance, a 1 percent increase of the weekly returns of one ton copper directs to a decrease of 4.5 percent of the stock returns for the sample firms in the period 1994-2001. This is in accordance with the negative relationship between the monthly movement of copper prices and stock returns found by Bailey & Chan (1993). However, on average, they obtain higher copper price exposure coefficients.

(35)

the factors of production (i.e. the rising oil price is absorbed by other factors of production such as labour, material or capital). Another reason may possibly be due to the relevance of crude oil as an input/output factor (Bartram, 2005). To see this, firms in the chemicals industry can be expected to have a higher oil derivative usage compared to firms in the construction industry since oil price movements have a larger impact on operations for chemical firms as opposed to construction firms. Based on this last notion, I analyze the commodity price exposure across different industries.

Table 12 presents the copper price, crude oil price and aluminum price exposures across six different industries estimated with Eq. (2) and Eq. (3) for the full sample period. Interestingly, crude oil exposure appears to be prevalent across industries. This can be due to the fact that crude oil is an important input/output for activities of many firms and that the usage of crude oil is evident in many different by-products such as toothpaste and packaging. An industry specific effect can be identified for crude oil, yet the number of exposures remains small across all industries. Various firms in the chemical industry, and iron and metal industry exhibit a significant negative crude oil exposure whereas several firms from the machinery and construction industry, electrical and electronic equipment industry and transport equipment industry predominantly exhibit positive crude oil exposure. A plausible reason why only a few firms in these industries are exposed to commodity price movements is that they might be aware that commodity price volatility is an important source of risk and hence engage in risk management activities. Bartram et al. (2009) and Bodnar et al. (1995) show that firms use derivatives to offset commodity risk, and that usage is concentrated in a few industries only such as oil and gas, chemicals, metal, and mining.

5.2 Commodity price exposure: Before and after the 1997 Asian crisis

(36)

Table 12.

Commodity price exposure 1994-2001

(See note)

1994-2001

Aluminum Copper Crude oil

- + +/- - + +/- - + +/-

Machinery and Construction 2 (11.8%) 1 (5.9%) 3 (17.6%) 1(5.9%) 0 (0.0%) 1 (5.9%) 0 (0.0%) 5 (29.4%) 5 (29.4%) 18 Electrical and Electronical Equipment 2 (10.5%) 1 (5.3%) 3 (15.8%) 0 (0.0%) 1 (5.3%) 1 (5.3%) 1 (5.3%) 5 (26.3%) 6 (31.6%) 20 Iron & Metal products 0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (5.9%) 1 (0.0%) 1 (5.9%) 2 (11.8%) 0 (0.0%) 2 (11.8%) 18 Transport Equipment 1 (6.7%) 0 (0.0%) 1 (6.7%) 0 (0.0%) 1 (6.7%) 1 (6.7%) 1 (6.7%) 6 (40.0%) 7 (46.7%) 16 Chemicals 1 (4.3%) 4 (17.4%) 5 (21.7%) 0 (0.0%) 2 (8.7%) 2 (8.7%) 3 (13.0%) 0 (0.0%) 3 (13.0%) 24 Food & Beverages 3 (27.3%) 0 (0.0%) 3 (27.3%) 4 (36.4%) 0 (0.0%) 4 (36.4%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 11

Note: The table reports the percentage of nonfinancial South Korean firms that show a significant commodity price exposure with regard to aluminum, copper and crude oil for different

industries (5% level). The three columns refer to the number and percentages of negative, positive and all exposures respectively. ‟ represent the total number of firms included in

the regression.

Commodity price hedging: A chaebol perspective

by

Nils de Jong

Commodity price hedging: A chaebol perspective

Table of Contents

1 Introduction

2 Literature review

3. Methodology

4. Data description

5. Empirical results first-step regression