The impact of foreign liabilities on small firms: Firm-level evidence from the Korean crisis

Journal of International Economics 97 (2015) 209–230

Contents lists available at ScienceDirect

Journal of International Economics journal homepage: www.elsevier.com/locate/jie

The impact of foreign liabilities on small firms: Firm-level evidence from the Korean crisis☆ Yun Jung Kim a, Linda L. Tesar b,c, Jing Zhang d,⁎ a

Sogang University, South Korea University of Michigan, USA NBER, USA d Federal Reserve Bank of Chicago, USA b c

a r t i c l e

i n f o

Article history: Received 25 April 2014 Received in revised form 27 May 2015 Accepted 27 May 2015 Available online 1 June 2015 JEL classification: F32 F34 E44

a b s t r a c t Using Korean firm-level data on publicly-listed and privately-held firms together with firm exit data, we find strong evidence that holdings of foreign-currency denominated debt negatively affected the economic performance of small firms during the 1997–98 crisis. The large exchange rate depreciation that occurred during the crisis resulted in a decline in net worth for firms with foreign-currency denominated debt on their balance sheets. Small firms with more short-term foreign debt were more likely to declare bankruptcy. Conditional on surviving the crisis, small firms that had more short-term foreign debt experienced larger declines in sales. The exit (bankruptcy) margin accounts for a large fraction of small firms' adjustment during the crisis. © 2015 Elsevier B.V. All rights reserved.

Keywords: Financial crisis Firm-level data Balance-sheet effects Korean economy

1. Introduction The sequence of events experienced by an emerging market undergoing a financial crisis is now all-too-familiar. Rapid economic growth and financial market liberalization encourage capital inflow, contributing to an overvalued exchange rate and increased reliance on foreign credit, usually denominated in US dollars. At some point, when economic growth and exports slow, the economy tips into crisis. The exchange rate collapses, capital flows reverse and firms find themselves unable to meet their debt requirements. Firms, and in some cases governments, become insolvent. Those deemed “too big to fail” may receive bailouts; others slash employment, declare bankruptcy or are sold to foreign owners.

☆ The authors would like to thank Chris House, Andrei Levchenko, Kathryn Dominguez, Amy Dittmar, and participants of seminars at the University of Michigan, the Australian National University, Sciences Po and the 2013 Economics Joint Conference in Korea. This work was supported by the National Research Foundation of Korea grant funded by the Korean Government (NRF-2013S1A5A8023475). ⁎ Corresponding author. E-mail addresses: [email protected] (Y.J. Kim), [email protected] (L.L. Tesar), [email protected] (J. Zhang).

http://dx.doi.org/10.1016/j.jinteco.2015.05.006 0022-1996/© 2015 Elsevier B.V. All rights reserved.

While the general anatomy of crises has been well documented,1 the channels through which a financial crisis translates into a real economic contraction at the microeconomic level are less well understood. A wide range of macroeconomic models predict that a depreciation of the exchange rate is expansionary by making exports more competitive (an export-expansion effect). On the other hand, models with financial frictions predict that an exchange rate depreciation is contractionary if firms have foreign-currency denominated liabilities on their balance sheets (a balance-sheet effect) (see, for example, Krugman, 1999; Céspedes et al., 2004; Feldstein, 1999). In general, the empirical literature has found ample evidence of the export-expansion effect but limited evidence of the balance-sheet effect. The fact that the evidence has pointed to a positive export-expansion effect has made it difficult to explain the steep output drop that occurs following balance of payments crises in small open economies. Our paper helps resolve this difficulty: in the case of Korea, the balance-sheet effect is found to be important for small, non-exporting firms that enter the crisis with short-term foreign-currency denominated debt. We use a detailed database that includes more than 4000 Korean firms to study the impact of the 1997–1998 Korean financial and currency

1

See for example Corsetti et al. (1999).

210

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

crisis on firm performance. Our database has a number of advantages relative to those used in previous studies. First, it includes information on publicly-listed as well as privately-held firms. This enables us to study the impact of the crisis on small firms, where the balance-sheet effects of holding unhedged, foreign-currency denominated debt were particularly acute. Second, the database contains firm-level information on export status, holdings of foreign debt, and total indebtedness together with a host of other variables. This allows us to condition on a number of firm-specific characteristics so that we can partially control for bias due to endogeneity. Third, the database provides information about firm exit during the crisis. We exploit heterogeneity across firms to examine whether exposure to foreign debt is critical for explaining firm performance and firm exit during the Korean crisis. Our empirical strategy is to regress measures of firm performance during the crisis on the pre-crisis foreign debt ratio. Using the lag in foreign debt holdings partially alleviates concerns about potential endogeneity between foreign debt holdings and firm performance. We further control for firm characteristics that are potentially related to foreign debt holdings and firm performance, including the export/sales ratio, size, age, an industry dummy, a chaebol dummy, leverage, and the short-term debt ratio.2 Finally, to the extent that unobserved omitted variables operate in a time-invariant fashion, the results from a regression in the pre-crisis period can serve as a benchmark for assessing the impact of the large exchange rate deprecation that occurred during the crisis. Our analysis yields three key findings on the magnitude and significance of balance-sheet effects. First, we find a negative relationship between foreign debt and firm net worth. This relationship is significant only during the crisis. Second, conditional on survival, foreign debt exposure has a negative effect on firm performance, particularly for small firms. Again, the relationship between firm size, foreign debt and firm performance is significant only during the crisis. To illustrate the magnitude of this effect, consider a firm at the 10th percentile of the size distribution (with size measured by real assets). For this firm, a 10% increase in its net short-term foreign debt to net worth ratio prior to the crisis is associated with a 1.7% reduction in real sales growth during the crisis. Third, we find that foreign debt holdings are a significant predictor of small firms' exit during the crises. For a firm at the 10th percentile of the size distribution, a 10% increase in its net short-term foreign debt ratio is associated with an increase of 1.3% in the probability of exit. Firm exit accounts for nearly 24% of the decline in aggregate sales in the peak year of the crisis. Our finding of a significant, negative balance-sheet effect, particularly for small firms, stands in contrast to most of the empirical work in this area, which reports either no effect or a positive balance-sheet effect.3 There are several explanations for the difference in findings. Most firm-level studies on emerging markets focus on publicly-listed firms. Publicly-listed firms tend to be large and are more likely to be exporters. Our data allows us to include small, privately-held firms where balancesheet effects turn out to be particularly important. Our specification also includes an interaction between firm size and financial variables. By allowing the balance-sheet effect to vary across firm size,4 we are able to uncover the strong balance-sheet effect on small firms. Finally, most previous analyses study only surviving firms. Analysis of exit rates underscores the devastating impact of the crisis on small firms that had foreign liabilities prior to the crisis.

2 Chaebols are South Korean conglomerates composed of many companies clustered around one parent company. 3 Benavente et al. (2003), Bleakley and Cowan (2008, 2009), Bonomo et al. (2003), Forbes (2002), and Luengnaruemitchai (2003) find either a positive balance sheet effect or no balance sheet effect. In contrast, Aguiar (2005), Carranza et al. (2003), Echeverrya et al. (2003), Gilchrist and Sim (2007) and Pratap et al. (2003) find some evidence of a negative balance sheet effect. 4 An interpretation of the firm size result is that size is a proxy for access to financial markets (see Gertler and Gilchrist, 1994).

Our empirical evidence identifies a strong export-expansion effect: exporters experience smaller declines in sales growth during the crisis than non-exporters. While in principle exports provide a natural hedge against the negative effects of an exchange rate depreciation and could provide a counter-weight to the negative balance-sheet effect, many firms do not export and therefore do not benefit from this channel. Our data suggest that only 30% of Korean firms that carried foreign currency debt on their balance sheets at the time of the crisis were exporters. Moreover, for the smallest quartile of firms, 90% of firms with foreign debt holdings were non-exporters. Therefore, a significant fraction of the population of Korean firms—importantly, most small firms—entered the crisis with exposure to balance-sheet risk with no potentially offsetting benefits from an improvement in export competitiveness. Paradoxically, we find that for large firms, exposure to foreign debt is positively associated with sales growth during the crisis. Similar results have been documented in other studies that focus on large, publiclylisted firms (see, for example, Bleakley and Cowan, 2008, 2009; Benavente et al., 2003; Bonomo et al., 2003; Forbes, 2002). We also find a negative association between foreign debt and exit during the crisis for large firms. The exact interpretation of these findings is unclear. One possibility is that large firms, like publicly-listed firms, are more likely to hedge exchange rate risk and are more likely to have access to other means of financing during the crisis. Unfortunately, we do not have direct evidence on the hedging behavior of firms or their holdings of other assets. It is reasonable to assume, however, that this measurement bias is less severe for small firms than for large firms; hence the balance-sheet effects we identify are stronger for small firms. The paper is organized as follows. Section 2 provides a brief macroeconomic overview of the Korean financial crisis and introduces the balance-sheet mechanism. The dataset used in our empirical analysis is discussed in Section 3. Section 4 presents empirical findings on the balance-sheet effects before and during the crisis for both the intensive and extensive margins. Robustness checks are performed in Section 5, and Section 6 concludes.

2. Macroeconomic dynamics of the Korean financial crisis Prior to the Asian financial crisis, South Korea was one of the fastest growing economies in the world, with sustained high real GDP growth rates for more than two decades. Beginning in late 1997, the Korean economy entered a severe economic contraction. Some indicators of the magnitude of the crisis are illustrated in Fig. 1a, which shows real GDP, consumption, investment and total employment normalized to their 1997 values.5 The declines were big: from peak to trough real GDP declined 7%, real consumption fell 14%, real investment fell 35%, and employment dropped 5%. During the crisis, the current account displayed a sudden reversal of over 15 percentage points, shifting from a negative balance of 4% of GDP to a positive 12% of GDP (Fig. 1b). While the crisis was deep, it was also mercifully brief. By 1999 real GDP and consumption returned to levels above their precrisis values. During the boom years, Korean firms and households dramatically increased their reliance on credit. Between 1995 and 1997, total private credit as a share of GDP increased from 104% of GDP to almost 120% of GDP (see Fig. 1d). Much of the credit expansion took the form of borrowing from abroad. Fig. 1c shows that external debt peaked at the end of 1997 at 60% of GDP, with over a third of total borrowing with maturities of one year or less. The declines in both total private credit and external debt as shares of GDP from 1997 to 2000 illustrate the dramatic deleveraging that occurred in Korea in the aftermath of the crisis.

5

The plot shows annualized data—the crisis hit in the fourth quarter of 1997.

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

(b) Current Account/ GDP

(a) Major Real Macro Variables (annual, 1997 level normalized to 1)

(%, annual) 15

1.4 GDP

C

I

N

1.2

10

1

5

0.8

0

0.6 1994

211

1995

1996

1997

1998

1999

2000

2001

2002

-5 1994

1995

1996

1997

1998

1999

2000

2001

2002

(d) Total Private Credit / GDP

(c) External Debt / GDP

(%, annual)

(%, annual) 120

80 Total 60

ST

110

40 100 20 0 1994

1995

1996

1997

1998

1999

2000

2001

2002

90 1994

1996

1997

1998

1999

2000

2001

2002

(f) Money Market Rate

(e) Nominal Exchange Rate

(%, monthly)

(KRW/USD, monthly) 1800

1995

30

1600 1400

20

1200 1000

1994 10

800 1994 1995 1996 1997 1998 1999 2000 2001 2002

0 1994 1995 1996 1997 1998 1999 2000 2001 2002

Fig. 1. Aggregate data. Note: The data source is Korea National Statistical Office.

Fig. 1e and f show the dynamics of two key prices: the nominal exchange rate (Korean won relative to the US dollar) and the nominal interest rate (the monthly money market rate). Prior to the financial crisis, South Korea had a long history of actively managing its exchange rate, effectively pegging the won to a basket of foreign currencies. As shown in Fig. 1e the nominal exchange rate depreciated by almost 100% during the last weeks of 1997, peaking in January 1998. Thereafter, the won fully floated against the dollar. It appears that there was significant overshooting of the Korean won—between late-1997 and mid1998 the won appreciated relative to the dollar although it did not return to its pre-crisis level. The short-term interest rate (Fig. 1f) also shot up during the crisis, increasing from its pre-crisis range of 10–15% to a peak of 25.6% in January 1998. The severity of the crisis has been attributed to high rates of leverage in the economy, particularly in the form of external debt, coupled with a sudden, unanticipated (and therefore unhedged) exchange rate depreciation. A wide range of macroeconomic models predict that a depreciation of the exchange rate will be expansionary by making exports more competitive. Exporting firms, all else equal, should benefit from an exchange rate deprecation. On the other hand, models with financial frictions predict that if the depreciation occurs when firms are holding significant foreign-currency denominated liabilities, a negative balance-sheet effect may outweigh the export-expansion effect. These models have three key predictions with respect to the balance-sheet effects. In response to an unexpected currency crisis, all else equal, firms with larger fractions of foreign currency denominated debt will

(i) experience larger declines in net worth; (ii) contract more and have worse performance; (iii) exit with higher likelihood.6 Despite the large literature on this topic, there has been little microeconomic evidence to support the connection between financial variables and real economic activity. Using the unique episode of Korean 1997 financial crisis, our paper studies empirically how both the decision to exit and firm performance were affected by the extent of foreign currency denominated debt on firm balance sheets, particularly for small firms. 3. Description of firm-level data Our analysis is based on data from the Korea Information Service, Inc. (KIS), a provider of financial and corporate data for Korean firms. The underlying source of the data is the annual financial statements of all Korean firms with assets over 7 billion won.7 The KIS removes liquidated firms from the dataset, and therefore the main dataset contains only surviving firms. We obtained additional information on liquidated firms by special request from the KIS. 6 The appendix provides a simple model to illustrate the mechanisms of the balancesheet model. 7 Firms with assets of 7 billion won or more are required by the Act on External Audit of Joint-Stock Corporations to report audited financial statements to the Financial Supervisory Commission, which is then compiled by the KIS. Some firms with assets less than 7 billion won voluntarily report their financial statements and show up in the dataset.

212

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

Table 1 Summary statistics. Surviving firms

Liquidated firms

All firms 1994

1995

1996

1997

2925 44.1

3671 39.6

3955 38.6

4597 37.3

3

Number of firms Fraction of firms with foreign debt (%) Fraction of exporters (%)

20.5

17.5

16.3

4 5 6

Mean age Mean real assets Median real assets

16 100 17

15 98 14

7 8 9

Median net worth growth (%) Median sales growth (%) Median investment/capital stock−1 (%) Median employment growth (%) Median personnel costs growth (%) Median raw material costs growth (%)

2.5 15.4 18.2

1 2

10 11 12

13 14 15 16 17

Mean leverage ratio (%) Mean ST debt ratio (%) Mean net foreign debt ratio (%) Mean net ST foreign debt ratio (%) Mean export/sales (%)

Publicly-listed firms 1998

1999

1994

1995

1996

1997

5067 34.0

5627 29.5

850 66.0

928 64.2

944 63.7

996 61.4

14.0

13.6

12.4

31.1

28.7

26.6

15 103 13

14 104 12

14 91 9

14 94 9

21 253 41

20 289 39

1.8 14.0 19.9

2.1 10.7 21.1

1.8 7.1 16.7

1.6 −9.5 10.6

7.9 20.7 16.5

2.9 15.3 18.7

1.4 14.4

0.9 13.2

0.0 11.5

0.0 6.2

−8.3 −11.9

1.9 18.8

15.1

14.8

10.7

6.7

−10.3

72.9 21.1 13.4 4.2

73.4 21.3 12.9 3.7

73.2 21.2 14.3 4.1

73.2 20.6 19.8 4.6

7.2

5.9

5.5

4.8

1998

1999

1995

1996

1997

1998

1999

1014 59.1

1020 53.9

71 22.5

100 16.0

218 25.7

206 28.2

57 24.6

22.4

21.6

19.4

12.7

5.0

10.6

5.8

5.3

21 320 41

20 357 39

20 340 32

20 391 33

11 17 13

9 19 13

10 20 13

12 25 12

12 15 11

2.8 14.4 20.6

2.7 9.3 20.7

1.8 7.5 17.7

2.7 −7.1 14.8

12.3 17.2 14.8

0.8 11.9

−0.8 18.5

−0.2 5.6

−1.3 5.7

−3.3 −29.7

−0.6 15.2

0.0 13.8

−0.8 10.7

−2.6 5.9

−12.1 −10.8

3.5 17.6

20.3

15.2

15.3

9.8

6.7

−7.2

16.7

68.3 20.3 13.0 3.6

65.4 18.7 10.0 3.1

69.4 19.1 15.6 6.3

69.7 19.4 15.6 6.1

69.2 19.7 18.2 6.9

69.3 19.5 23.8 7.6

64.1 18.2 16.0 5.7

56.7 14.4 10.3 4.7

92.8 39.8 10.0 2.3

98.3 40.1 8.5 1.8

96.2 36.9 23.2 3.9

104.0 45.1 14.6 3.3

112.7 45.3 23.3 6.7

5.2

4.5

11.0

10.0

9.2

7.6

8.8

7.4

6.6

3.4

3.5

3.0

2.0

Note: Real assets are in billion 1994 won. All growth rates are real. The leverage ratio is defined as total liabilities over total assets. The short-term debt ratio is defined as the amount of debt due within twelve months divided by total assets. The net foreign debt ratio is defined as net foreign debt as a share of net worth. The net short-term foreign debt ratio is defined as net short-term foreign debt over net worth. Characteristic statistics of exited firms are reported for the year preceding the liquidation.

The KIS data have several advantages over the data that have been employed in earlier studies of financial crises in emerging markets.8 First, the KIS data include firms that are not listed on the Korean stock exchange. The KIS data reveal that publicly-listed firms are only a fraction of the population of Korean firms and they provide a skewed portrait of the impact of the crisis at the micro level. The KIS data also provide information on foreign debt versus domestic debt,9 as well as the maturity structure of the debt. The database contains firm-level information on exports, allowing us to disentangle the export-expansion effect from the balance-sheet effect of an exchange rate depreciation. Finally, the merged database allows us to study firm exit, a margin of adjustment during the Korean crisis that has not heretofore been studied. We focus on the 1994–1999 sample period to capture the effects of the financial crisis. We exclude firms in the financial sector.10 In order to limit the influence of outliers, we eliminate observations in the top and bottom 1% of the sample in terms of the sales growth rate and foreign debt ratios.11 When firms are sorted by industry, about 58% of firms are in the manufacturing sector, 15% in wholesale, retail trade and 8 Examples, among many others, include Aguiar (2005), Bleakley and Cowan (2008), Borensztein and Lee (2002), Forbes (2002), Gilchrist and Sim (2007), Kalemli-Ozcan et al. (2010), and Martínez and Werner (2002). All these papers focus on publicly-listed firms. Bleakley and Cowan (2008) have no information on export status and Forbes (2002) uses total debt statistics instead of foreign debt. All these papers, except KalemliOzcan et al. (2010), have no exit information. In Kalemli-Ozcan et al. (2010) firms rarely exit, so the exit margin plays a limited role in their study. 9 The KIS does not provide the currency denomination of foreign debt. However, other sources indicate that the majority of foreign borrowing was denominated in US dollars. According to Bae and Kwon (2010), prior to the crisis 96% of foreign debt of publiclylisted firms was in US dollars, 3% in yen, and 1% in other currencies. 10 The balance sheets of financial firms are quite different from those of real-sector firms. It is also hard to measure the performance of financial firms. An additional challenge is the substantial regulatory changes that took place in this sector in Korea around the time of the crisis. 11 Table 1 computes the statistics based on the observations included in the sales growth regressions of specification 1 and 2 as described later in Section 4. For the regression analysis, we first compute the dependent and independent variables. We then replace the values of the top and bottom 1% outliers of each dependent variable and foreign debt ratios with missing values. The observations with missing values in the dependent variable or any independent variables in the specifications are automatically dropped in the regressions.

transportation, 12% in construction and utility, and the remaining 15% provide other services. These industry shares are fairly constant over the 1994–1999 period. Table 1 provides summary statistics for both surviving firms and liquidated firms. The sample of firms starts with a sample size of 2925 and increases over the 1994–1999 period (line 1). The increase in the sample size over time is not surprising given that the cutoff for inclusion in the database (7 billion won) is fixed in nominal terms; as the economy grows and there is inflation, the number of firms above this cutoff will obviously increase. Note that small firms in our sample are not small in the universe of Korean firms. The cutoff firms have an asset equivalent to 8.3 million US dollars at the end of 1996. The number of firms with foreign debt exposure (line 2) is large: about 39% of the full sample of firms carried foreign-currency denominated debt on their balance sheets in 1996. Over 16% of firms reported exports in 1996 (line 3). The mean age of firms (Table 1, line 4) is 14 to 16 years. In the first year, the median level of total assets (line 6) is 17 billion won, twice the cutoff level for inclusion in the database. The mean level of real assets (line 5) is dramatically larger at 100 billion won, suggesting that the full sample covers many smaller firms. As we show below, inclusion of relatively small firms is critical for identifying the balance-sheet effect on firm performance during the crisis. The median growth rate of real net worth is around 2% prior to the crisis, and declines to 1.7% in 1997–98 (line 7). The focus of our analysis will be firm performance during the crisis as measured by sales growth rates. The annual median real sales growth rate (lines 8) is in the 10–15% range in the pre-crisis period. The crisis occurred in late 1997, and median real sales growth drops off to 7.1% that year and then plummets to −9.5% in 1998. We also consider alternative performance measures using the investment/capital ratio, employment growth rate, growth rates of personnel costs and raw material costs, which are discussed in the robustness checks. These variables, reported in line 9–12, depict similar dynamics as sales growth. Firm-level financial statistics are shown in lines 13 to 16. The leverage ratio (line 13) is defined as total liabilities over total assets. The short-term debt ratio (line 14) is short-term debt over total assets.

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

213

External Debt (billion dollar, end of year) 120 Aggregate Banking Sector

LT

Firms in KIS Data

100

ST 80 60 40 20 0

1994

1995

1996

1997

1998

1999

2000

2001

2002

Fig. 2. Aggregate and firm-level debt data. Note: Short-term debt has original maturity equal to or less than one year. The aggregate debt statistics come from Korea National Statistical Office, and the firm-level debt statistics come from the KIS-VALUE dataset.

Short-term refers to debt due within the next twelve months. The net foreign debt ratio (line 15) is computed as the ratio of net foreign debt to net worth. Net foreign debt is foreign currency denominated debt minus foreign currency denominated assets and net worth is defined as total assets minus total liabilities.12 The net short-term foreign debt ratio (line 16) is the ratio of net short-term foreign debt to net worth.13 The mean leverage ratio declines after the crisis from 73% in 1994–1997 to 65% in 1999. The short-term debt ratio is relatively constant over the period of 1994–1999 at around 20%. The net foreign debt ratio is about 14% before the crisis and rises to 19.8% in 1997 in part due to the exchange rate depreciation. The net short-term foreign debt ratio is about 4.1% before the crisis and rises to 4.6% in 1997. For those firms reporting foreign liabilities, the mean and median net foreign debt ratios were 37 and 17% in 1996, respectively. Fig. 2 compares the level of foreign currency debt of the banking sector and the sum of foreign currency debt of the firms in our sample. In both cases debt is decomposed into short-term and long-term debt, where short-term debt is defined as debt with original maturity of one year or less. External debt of both banks and private firms in our sample increased in the years preceding the crisis, with short-term debt accounting for roughly half of all external liabilities. This pattern is not surprising because the majority of foreign debt holdings by Korean firms are channeled through the domestic banking sector. Previous analyses of emerging market crises suggest that exports may have provided firms with a natural hedge for foreign currency exposure—a depreciating currency will increase the cost of dollardenominated debt service, but will increase the firm's competitiveness in foreign markets.14 Firm exports as a share of total sales are reported in line 17 of Table 1. The mean export to sales ratio is around 5% in our sample period. Conditional on exporting, the average export to sales ratio is around 35%. The middle panel of Table 1 reports summary statistics for publiclylisted firms, which account for around 20% of the full sample.15 Publicly12 For foreign currency denominated assets, the KIS data provide information only on cash holdings in foreign currency. The database does not provide information on other forms of foreign assets. 13 In constructing the net foreign debt to net worth ratios, we eliminate observations with negative net worth. In the regressions, we eliminate the top and bottom 1% of the foreign debt ratios to limit the influence of outliers. 14 See Aguiar (2005) for the evidence from Mexico, Bleakley and Cowan (2008) for five Latin American countries, and Luengnaruemitchai (2003) for six East Asian countries. 15 Publicly-listed firms account for 74% of sales and 41% of employment in our 1996 sample.

listed firms are older and larger than the average firm in the full sample. Publicly-listed firms have similar performance dynamics as the full sample. They have somewhat lower leverage ratios and short-term debt ratios. They are more exposed to foreign-currency denominated debt. Firms holding foreign-currency denominated debt constitute about 64% of the sample of publicly-listed firms but only 39% of the full sample in 1996. Although publicly-listed firms tend to hold more foreign debt, the foreign exposure of privately-held firms is not trivial. More than 900 privately-held firms in the full-sample are foreign debtors in 1996. Including those firms into the analysis will prove to be important for our results. The average net foreign debt ratio is also larger for publicly-listed firms at 18% in 1996 while it is 14% in the full sample. The net shortterm foreign debt ratio is about 7% in the sample of publicly-listed firms but 4% in the full sample. Conditional on having dollar debt, on the other hand, the average net foreign debt ratio is smaller for publicly-listed firms at 29% in 1996, while it is 37% for the full sample, suggesting the full sample has a more skewed distribution of the net foreign debt ratios than the sample of publicly-listed firms. Publicly-listed firms are more likely to be exporters. The fraction of firms that are exporters is about 27% among publicly-listed firms, while only 16% among the full sample in 1996. Conditional on having positive exports, the mean export/sales ratio is similar across these two samples.16 An important issue is the extent to which our sample of firms is representative of the aggregate economy. While our empirical work will exploit heterogeneity across firms, our results could be viewed with suspicion if our sample of firms exhibits aggregate sales behavior during the crisis that is dramatically different from the dynamics of aggregate

16 One group of Korean firms that has received a great deal of attention is the subset of firms belonging to chaebols. As the literature has emphasized, membership in a chaebol can provide insurance to firms through interlocking contracts and financial linkages. See Borensztein and Lee (2002), Lee et al. (2000), and Min (2007). Our dataset includes roughly 250 firms that are part of the top 30 chaebols. Their characteristics tend to be similar to those of publicly-listed firms with four exceptions. First, the size of a chaebol firm, as measured by mean real assets, is more than twice the size of publicly-listed firms, and about seven to twelve times larger than the mean firm in the full sample. Second, the chaebols tend to have larger sales growth rates but lower profit rates than the publicly-listed firms. Third, the chaebols have much larger leverage ratios and greater exposure to foreign debt than the publicly-listed firms. Finally, the chaebols have smaller export/sales ratios than the publicly-listed firms. We include a chaebol dummy in our cross-section analysis to test for the role of network linkages on firm performance. No chaebols exited from the sample prior to the financial crisis.

214

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

(b) Real GDP and Firm Sales Growth

(a) Sum of Firm Sales / GDP

(annual, %)

(annual) 1.4

25.0 Real GDP Growth

All Firms

1.3

20.0

Public Firms

1.2

Median Real Sales Growth

15.0

1.1 10.0 1 5.0 0.9 0.0

0.8

-5.0

0.7 0.6

1994

1995

1996

1997

1998

1999

2000

2001

2002

-10.0

1994

1995

1996

1997

1998

1999

2000

2001

2002

Fig. 3. Comparison of firm sales and GDP. Note: The GDP data come from Korea National Statistical Office, and the firm sales data are from the KIS-VALUE dataset.

Table 2 Summary statistics by firm size. 1995

1996

1997

1998

1999

Fraction of firms with foreign debt Below median Above median

22.8 56.3

21.4 55.8

19.2 55.3

15.3 52.7

12.9 45.8

Mean foreign debt/NW of firms with foreign debt Below median Above median

37.9 30.6

42.0 35.3

63.7 49.6

44.9 36.5

50.3 29.2

Fraction of exporters Below median Above median

11.3 23.7

10.2 22.5

8.6 19.5

9.0 18.2

8.3 16.4

Mean export/sales of exporters Below median Above median

30.8 35.0

33.3 34.2

31.7 35.3

28.6 42.7

29.7 39.5

Aggregate sales growth (1) = (2) + (3) Below median Above median

18.6 15.0 18.8

12.8 15.3 12.7

11.2 −5.9 4.0 −13.3 11.6 −5.6

13.6 49.3 12.0

Sales growth accounted by survivors (2) Below median Above median

18.9 17.6 19.0

13.3 18.1 13.0

12.3 10.5 12.4

−4.5 −7.4 −4.4

13.8 50.3 12.1

Sales growth accounted by exit (3) Below median Above median

−0.3 −2.6 −0.2

−0.4 −2.9 −0.3

−1.1 −6.6 −0.8

−1.4 −0.2 −5.9 −1.0 −1.2 −0.2

% of exit margin in aggregate sales growth (3)/(1) Below median Above median

−1.9

−3.2

−9.8

23.7 −1.4

−17.6 −18.9 −166.0 −1.0 −1.9 −6.5

44.1 −2.1 26.5 −1.1

Note: Firms below median are firms with assets below median size, and firms above median are firms with assets above median size each year. All statistics are in percentage terms.

economic activity. To address this issue, Fig. 3a shows the sum of firm sales as a ratio of GDP. The top line is for all firms in our sample. The ratio is just under 1 in 1994 and increases to about 1.4 in 2000 as more firms are brought into the sample.17 The figure also shows the ratio for publicly-listed firms, which tops out at about 0.9. Fig. 3b compares the time series of real GDP growth over 1994–2002 to median

17 These numbers are smaller than the output/GDP ratio for the US economy. Based on BEA data, the ratio of gross output of all industries excluding the financial industry to GDP ranges from 1.64 to 1.7 between 1994 and 2007. Thus, firm coverage of the KIS database might be somewhat less complete than the BEA coverage.

real sales growth for our full sample of firms. Not surprisingly, there is more variation in sales growth than in GDP, but the shape of the two curves is similar. Both series pick up the dramatic fall in economic activity in 1998 and the recovery in 1999. This suggests that the patterns we see in firm-level data are consistent with aggregate macroeconomic dynamics. We now turn to liquidated firms in the sample, reported in the right panel of Table 1 (the statistics are for the year preceding firm exit). The KIS database includes a list of firms that submitted a notification of closing business to the court system and balance sheet information for these firms before their liquidation.18 The exit rate in our sample was around 2% in the pre-crisis years, and it doubled to 4.7% in 1997 and remained high at 4.1% in 1998. The exit rate dropped to 1% in 1999.19 It should be noted that no publiclylisted firms filed a notification of closing throughout the 1994–1999 period. No chaebol firms exited before the crisis, though some did exit during the crisis. A large fraction of firms liquidated in the crisis had foreign debt in the previous year (line 2): 26% in 1997 and 28% in 1998. Liquidated firms were less likely to be exporters than surviving firms (line 3). Comparing liquidated firms with surviving firms, we see that liquidated firms tend to be younger and smaller in size than the average firm (line 4–6). Before they exit, firm-level net worth growth rates are very low or negative (line 7). Prior to exit, liquidated firms carry substantially more debt, particularly short-term debt, relative to surviving firms (line 13–14). Firms liquidated prior to the crisis have smaller foreign debt ratios than surviving firms. But the average net foreign debt ratio of the firms liquidated in 1997 is 23% in 1996, much higher than 14% for surviving firms (line 15). Exiting firms tend to be concentrated in the construction and manufacturing sectors. So far we have presented the unconditional summary statistics of the dataset. We next show the summary statistics of key variables such as foreign debt, exports and sales growth, for firms above or below the median size in Table 2. These conditional statistics shed light on differential firms' behavior and performances across firm size. Looking at the top panel of Table 2, we find that large

18 The list of liquidated firms does not include reorganized firms or firms that were sold to a foreign company. Thus, our exit data underestimates the severity of bankruptcy in crisis. The dataset does not allow us to precisely track entry. Firms may appear in the database either because they are newly established or because they reach the 7 billion won criterion. 19 The exit rate in year t is computed as the number of firms that exited in year t divided by the sum of the number of surviving firms from year t − 1 to t and the number of firms that exited in year t.

percent

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

(b) Median Sales Growth

by Industry

by Firm Size

(c) Median Sales Growth by Export Status

30

30

30

20

20

20

10

10

10

0

0

0

-10

-20 1994

percent

(a) Median Sales Growth

215

Ind1 Ind2 Ind3 Ind4 Ind5 1995

-10

1996

1997

1998

1999

-20 1994

Group 1 - smallest Group 2 Group 3 Group 4 - largest 1995

1996

-10 Exporters Non-Exporters 1997

1998

(d) Median Sales Growth

(e) Median Sales Growth

by Leverage

by ST Debt Ratio

1999

-20 1994

30

20

20

20

10

10

10

0

0

0

-20 1994

1995

1996

-10

1997

1998

1999

-20 1994

Group 1 - smallest Group 2 Group 3 Group 4 - largest 1995

1996

1997

1998

1999

by Foreign Debt Ratio

30

Group 1 - smallest Group 2 Group 3 Group 4 - largest

1996

(f) Median Sales Growth

30

-10

1995

-10 Foreign Debt No Foreign Debt 1997

1998

1999

-20 1994

1995

1996

1997

1998

1999

Fig. 4. Sales growth of firms with varying characteristics. Note: Industry 1 is agriculture, forestry, fishing, and mining. Industry 2 is construction and utility. Industry 3 is manufacturing. Industry 4 is wholesale and retail trade and transportation. Industry 5 is other services. The data source is the KIS-VALUE dataset.

firms (above the median) are more likely to hold foreign debt than small firms (below median). In 1996, about 56% of large firms have the foreign debt exposure, while only 21% of small firms have the exposure. However, conditional on having foreign debt, small firms hold more net foreign debt relative to their net worth. Consistent with the empirical trade literature, we find that large firms are more likely to export and conditional on exporting, exports account for a large fraction of their sales. In the lower panel of Table 2, we decompose the decline in firm sales growth into the drop in sales of surviving firms (the survivor margin), and the drop due to firm exit (the exit margin) for the aggregate economy and the two groups of firms by size.20 Consider the change in firm sales of a particular group between year t and year t + 1. Some firms in year t continue in operation in year t + 1, and we refer to these firms as “surviving firms”. The remaining firms liquidate and exit, and we refer to them as “exiting firms”.21 The net sales growth equals the ratio of total sales of surviving firms in period t + 1 and total sales of both surviving and liquidated firms in period t. We decompose sales growth into

20 In this analysis, we abstract from the entry margin because the KIS dataset does not cover many entering firms. We doubt that the entry margin plays an important role during the financial crisis. 21 In this analysis, we ignore the contribution to total sales growth by firms that enter the database in period t + 1.

the survivor and exit margins as follows:

SalesSurviving;tþ1 SalesSurviving;tþ1 −SalesSurviving;t −1 ¼ SalesSurviving;t þ Salesexiting;t SalesSurviving;t þ Salesexiting;t −

Salesexiting;t ; SalesSurviving;t þ Salesexiting;t

where the left hand side is the net sales growth, the first term on the right hand side is the survivor margin and the second term is the exit margin. The survivor margin is the ratio of the change in total sales of surviving firms between year t and t + 1 and total sales in period t, and the exit margin is the ratio of total sales of exiting firms and total sales of all firms in year t. As the table shows, the contribution of the exit margin to aggregate sales growth is small prior to the crisis—about 3% of total sales growth in our sample. In the crisis years, however, the exit margin becomes substantially more important, accounting for 24% of the fall in aggregate sales growth in 1998. This reinforces the point that excluding exiting firms from the analysis will miss a significant part of the macroeconomic adjustment that occurred during the Korean crisis. The extensive margin impact on sales growth is much more significant for small firms, particularly during the crisis.

216

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

Table 3 Changes in net worth and working capital. Publicly-listed firms

Full sample Crisis NW Chaebol dummy Age

0.016⁎⁎ (0.008) −0.000⁎⁎⁎

Size

(0.000) −0.013⁎⁎⁎ (0.002)

Leverage ratio

−0.035⁎⁎⁎

ST debt ratio Export/sales ratio Foreign debt ratio

Observations R-squared

Pre-crisis WC 0.012 (0.020) 0.000 (0.000) −0.020⁎⁎⁎

NW 0.020⁎⁎ (0.009) −0.001⁎⁎⁎

(0.007)

(0.000) 0.000 (0.002)

−0.121⁎⁎

−0.053⁎⁎⁎

Crisis NW

−0.011 (0.014) 0.000 (0.000) 0.004 (0.005)

0.030⁎⁎ (0.014) −0.001⁎⁎ (0.000) −0.021⁎⁎⁎ (0.004)

(0.017)

(0.000) −0.009⁎⁎ (0.004)

−0.073⁎⁎ (0.036) 0.037 (0.046) 0.010 (0.020)

−0.169⁎⁎ (0.066) 0.299⁎⁎⁎

0.004 (0.034) −0.156⁎⁎⁎

(0.084) −0.028 (0.034)

(0.038) 0.008 (0.019)

−0.063 (0.040) 0.050 (0.049) −0.003 (0.021)

−0.021 (0.013)

−0.065⁎⁎ (0.028)

−0.010 (0.012)

−0.003 (0.022)

(0.012)

(0.031) −0.004 (0.018)

(0.014) −0.011 (0.010)

−0.046⁎ (0.025) 0.024 (0.029) −0.018 (0.027)

−0.012⁎⁎⁎ (0.005)

−0.016⁎ (0.009)

−0.002 (0.005)

−0.008 (0.009)

(0.013) −0.007 (0.014) 0.027⁎⁎

4104 0.113

(0.049) 0.130⁎⁎⁎

4140 0.054

(0.015) −0.056⁎⁎⁎

2997 0.079

Pre-crisis

WC

3006 0.039

958 0.197

WC 0.057 (0.042) −0.001 (0.001) −0.028⁎

970 0.120

NW

WC

0.045⁎⁎⁎ (0.016) −0.001⁎

−0.004 (0.021) −0.001 (0.000) −0.003 (0.006)

851 0.120

851 0.075

Note: The dependent variables are indicated in the third row. NW refers to Δ(Net Worth)/Assets−1 and WC refers to Δ(Working Capital)/Assets−1. The dependent variable is the change in NW and WC between 1996 and 1997 relative to total assets in 1996 for the crisis regressions, and the change between 1994 and 1995 relative to total assets in 1994 for the pre-crisis regressions. The independent variables are for year 1996 in the crisis regressions and for year 1994 in the pre-crisis regressions. Robust standard errors are reported in parentheses. All regressions include industry dummies at the two-digit level. ⁎⁎⁎ Denotes a p-value less than 1%. ⁎⁎ Denotes a p-value less than 5%. ⁎ Denotes a p-value less than 10%.

4. Empirical results Before turning to the regression analysis, we first plot the time series of median sales growth for different groups of firms. We classify firms into groups according to their characteristics in each year. Fig. 4 shows median sales growth for firms by industry, firm size, export status, leverage, short-term debt and foreign debt ratios. The overwhelming message of Fig. 4 is that the economic contraction was a macroeconomic phenomenon. While there are some differences across firms—for example, sales of non-exporters contracted more sharply than exporters, and sales of the other services industry had the deepest fall in 1998—virtually all sectors and all types of firms experienced a deep contraction in 1998 and a sharp recovery in 1999. There is no discernible difference in the sales growth rates prior to the crisis between firms with and without foreign debt holdings. Paradoxically, firms with no foreign debt experienced larger declines in sales growth during the crisis than firms with foreign debt. This unconditional pattern might explain why there is scarce empirical support for the balance-sheet effect at the microeconomic level, even though the impact of a currency crisis is large at the macroeconomic level. It also underscores the need to include interaction effects in the empirical analysis. We turn to the regression analysis to test the three key theoretical predictions on the balance-sheet effects: during an unexpected currency crisis, firms with larger exposure to foreign currency denominated debt (i) experience larger declines in net worth, (ii) have worse performance, and (iii) have higher likelihood of exit. We present the empirical results for the three predictions in the following three subsections. Our empirical strategy regresses measures of firm performance during the crisis on the pre-crisis foreign debt ratio. Using the lag in foreign debt holdings partially alleviates concerns about endogeneity between foreign debt holdings and firm performance. However, there may remain omitted variables that co-vary with both foreign debt and performance, which could bias our results. For example, a firm with low productivity might tend to borrow in foreign currency and would also have poor economic performance as measured by sales. We address the omitted-variable problem in two ways. First, we include firm

characteristics that are potentially related to foreign debt holdings and firm performance in the regression. Second, for unobserved characteristics such as firm productivity, we use the pre-crisis period as a benchmark for evaluating the crisis results. Referring back to our example above, if it is low productivity that is driving poor performance, then the relationship between foreign debt holdings and performance should hold in both the pre-crisis and crisis periods. If instead, as we find in our regression analysis, foreign debt only matters during the crisis, we have some confidence that this is because of the exchange rate shock working through the balance-sheet effect, and not a spurious relationship between foreign debt and productivity. 4.1. Changes in net worth following the crisis To investigate the direct impact of a currency crisis on firm net worth, we regress changes in net worth and working capital immediately following the exchange rate depreciation on pre-crisis firm characteristics:

ΔNET WORTHi ¼ α þ β FDi; −1 þ γ CHARi; −1 þ εi :

ð1Þ

Net worth is measured as total assets minus total liabilities and working capital as current assets minus current liabilities. Working capital therefore reflects the firm's liquid net worth. In the regression, we take the change in net worth and working capital over the interval end-1996 to end-1997 as a share of assets in 1996. Pre-crisis firm characteristics are variables observed in 1996. FD denotes the foreign debt ratio, and CHAR includes the log of firm real assets, age, chaebol status, leverage, short-term debt ratios, and export/sales ratios. All variables are in real Korean won. We include a two-digit industry dummy to control for industry-specific effects. To highlight the role of foreign debt during the crisis, we conduct the same exercise for a pre-crisis period by regressing changes in net worth from 1994 to 1995 on firm characteristics in 1994.

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

217

Table 4 Cross-section regressions for publicly-listed firms. Dependent variable Sales growth Chaebol dummy Age Size

Crisis 2

3

4

5

6

0.049 (0.039) 0.001 (0.001) −0.061⁎⁎⁎

0.058 (0.036) 0.001 (0.001) 0.002 (0.057)

0.047 (0.036) 0.001 (0.001) 0.000 (0.057)

0.067 (0.041) −0.001⁎⁎ (0.001) −0.025⁎⁎

0.074⁎ (0.042) −0.001⁎⁎ (0.001) −0.022 (0.034)

0.067 (0.043) −0.002⁎⁎ (0.001) −0.023 (0.034)

2.538 (1.704) −0.102 (0.068) −1.888 (1.895) 0.067 (0.073)

2.379 (1.664) −0.100 (0.066) −1.541 (1.810) 0.061 (0.072)

0.045 (0.066)

0.382 (1.225) −0.014 (0.050) −1.918 (1.824) 0.077 (0.074)

0.403 (1.230) −0.014 (0.050) −1.442 (1.985) 0.059 (0.081)

−0.016 (0.049)

−0.014 (0.051)

−0.010 (0.051)

0.010 (0.056)

0.701 (0.619) −0.028 (0.024)

(0.018) Leverage ratio

0.061 (0.095)

Size ∗ leverage ratio ST debt ratio

Pre-crisis

1

−0.287 (0.215)

Size ∗ ST debt ratio Export/sales ratio

0.216⁎⁎⁎ (0.061)

Foreign debt ratio

0.258 (0.166)

Size ∗ foreign debt ratio

0.218⁎⁎⁎ (0.061)

0.183⁎⁎⁎ (0.066)

0.511 (1.748) −0.010 (0.065)

−0.056 (0.121)

−3.307⁎ (1.959) 0.129⁎

ST foreign debt ratio Size ∗ ST foreign debt ratio

−0.630 (0.787) 0.022 (0.031) 0.044 (0.770) −0.003 (0.031)

(0.074) 1.667 (3.211) −0.046 (0.120)

LT foreign debt ratio Size ∗ LT foreign debt ratio Observations R-squared

(0.012)

944 0.188

944 0.191

934 0.224

850 0.171

850 0.177

845 0.176

Note: The dependent variable is firm real sales growth between 1997 and 1998 for the crisis regressions and real sales growth rate between 1995 and 1996 for the pre-crisis regressions. The independent variables are for year 1996 in the crisis regressions and for year 1994 in the pre-crisis regressions. Firms with negative net worth are excluded from the sample. Robust standard errors are reported in parentheses. All regressions include industry dummies at the two-digit level and the lagged sales growth rate. ⁎⁎⁎ Denotes a p-value less than 1%. ⁎⁎ Denotes a p-value less than 5%. ⁎ Denotes a p-value less than 10%.

Table 3 compares the results for the two samples of firms (all firms and publicly-listed firms) over the two periods (crisis and pre-crisis). For the full sample, we find that larger exposure to foreign debt is associated with significantly larger declines in net worth and working capital following the currency crisis, consistent with implications of the balance-sheet effect models. For publicly-listed firms, foreign debt holdings do not seem to affect changes in net worth but they do affect changes in working capital during the crisis. In the pre-crisis period, foreign debt has no effect on changes in net worth or working capital. 4.2. Surviving firms' performance during the crisis We next examine the impact of foreign debt holdings on sales growth. The general form for the cross-section regressions is shown in the following equation22: SALES GROWTHi ¼ α þ β FDi; −1 þ γ CHARi; −1 þ εi :

ð2Þ

22 Our goal is to account for the cross-sectional variation in firm performance during the crisis, and to relate this variation to firm-specific pre-crisis characteristics. An alternative would be to use a panel specification with firm fixed effects, and estimate how withinfirm variation in debt holdings and export sales affects variation in firm performances over time. In that case, the impact of the crisis would be estimated through an interaction of lagged firm characteristics with the crisis dummy. We do not pursue this strategy as the baseline estimation for three reasons. First, such a specification would answer a different, much narrower question: how does the crisis affect the relationship between debt holdings or export sales and sales growth within a firm? Second, firm fixed effects will soak up the explanatory power of interesting and informative time-invariant firm characteristics. Third, the short-time dimension of our dataset implies that we have limited time variation to exploit. We report the results from the panel analysis as a robustness check in Section 5.3.

The dependent variable is firm i's annual real sales growth and the independent variables are pre-existing firm characteristics. As the firm performance variable is likely to be serially correlated, we include a lagged dependent variable in all specifications. We perform the regression analysis for two time periods—the crisis period (characteristics in 1996 as explanators for the sales growth rate between 1997 and 1998) and the pre-crisis period (characteristics in 1994 as explanators for the sales growth rate between 1995 and 1996). We compare the regression results across the two periods and across two samples— publicly-listed firms and the full sample which includes smaller and privately-held firms. In the baseline specification, firm-level characteristics are the same as those in the net worth regressions. In the second specification, we also include interaction terms between firm size with the net foreign debt ratio, the leverage ratio, and the short-term debt ratio. Because the balance-sheet effect varies with a firm's financial access, which is correlated with firm size in the data,23 introducing the interaction terms helps uncover the balance-sheet effect on firms with limited financial access. In the third specification, we decompose the net foreign debt ratio by maturity to examine whether firms with varying foreign debt maturities have differential firm performance.

23 Fazzari et al. (1988) and Gertler and Hubbard (1988) show that small firms rely more heavily on bank finance, especially in the form of short-term debt, while large firms rely more on non-bank finance. They also show that net worth is more important in explaining investment behavior for small firms, an indication that small firms are more likely to face capital market frictions.

218

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

Table 5 Cross-section regressions for the full sample. Dependent variable Sales growth Chaebol dummy Age Size

Crisis 2

3

4

5

6

0.081⁎⁎⁎ (0.027) 0.000 (0.001) −0.043⁎⁎⁎

0.072⁎⁎⁎ (0.028) 0.001 (0.001) −0.004 (0.027)

0.067⁎⁎ (0.028) 0.001 (0.001) −0.005 (0.027)

0.052⁎⁎ (0.024) −0.002⁎⁎⁎ (0.001) −0.016⁎⁎

0.053⁎⁎ (0.025) −0.002⁎⁎⁎ (0.001) −0.022 (0.034)

0.060⁎⁎⁎ (0.023) −0.002⁎⁎⁎ (0.001) −0.025 (0.034)

0.750 (0.915) −0.031 (0.039) 2.431⁎⁎

0.834 (0.912) −0.035 (0.039) 2.239⁎⁎

0.017 (0.052)

(1.027) −0.103⁎⁎

(1.041) −0.094⁎⁎

(0.044)

(0.044)

0.295 (1.168) −0.012 (0.048) −1.739 (1.368) 0.077 (0.057)

0.209 (1.169) −0.008 (0.048) −1.489 (1.404) 0.066 (0.059)

−0.020 (0.029)

−0.018 (0.029)

−0.015 (0.029)

0.025 (0.021)

0.085 (0.446) −0.003 (0.019)

(0.007) Leverage ratio

0.007 (0.047)

Size ∗ leverage ratio ST debt ratio

0.039 (0.059)

Size ∗ ST debt ratio Export/sales ratio Foreign debt ratio

Pre-crisis

1

0.179⁎⁎⁎ (0.034) 0.065⁎⁎ (0.033)

Size ∗ foreign debt ratio

0.173⁎⁎⁎ (0.034)

0.179⁎⁎⁎ (0.034)

−0.621⁎⁎ (0.311) 0.029⁎⁎

(0.007)

0.073 (0.100)

(0.012) −1.647⁎⁎ (0.642) 0.068⁎⁎⁎

ST foreign debt ratio Size ∗ ST foreign debt ratio

(0.025) −0.864 (0.551) 0.042⁎ (0.022)

LT foreign debt ratio Size ∗ LT foreign debt ratio Observations R-squared

−0.256 (0.543) 0.010 (0.022) 0.082 (0.461) −0.003 (0.019)

3955 0.117

3955 0.121

3921 0.125

2925 0.090

2925 0.091

2911 0.092

Note: The dependent variable is firm real sales growth between 1997 and 1998 for the crisis regressions and real sales growth rate between 1995 and 1996 for the pre-crisis regressions. The independent variables are for year 1996 in the crisis regressions and for year 1994 in the pre-crisis regressions. Firms with negative net worth are excluded from the sample. Robust standard errors are reported in parentheses. All regressions include industry dummies at the two-digit level and the lagged sales growth rate. ⁎⁎⁎ Denotes a p-value less than 1%. ⁎⁎ Denotes a p-value less than 5%. ⁎ Denotes a p-value less than 10%.

4.2.1. Results for publicly-listed firms We begin with the results for the sample of publicly-listed firms in Table 4. Columns 1, 2 and 3 report the results for the crisis period, and columns 4, 5 and 6 report the results for the pre-crisis period. In the pre-crisis period, firm age and size are negatively related to firm performance (column 4). Firm size loses its significance when it is interacted with other financial variables (columns 5 and 6). The coefficients on chaebol status are positive but not statistically significant in general. There is no evidence of a significant relationship between the performance of publicly-listed firms in the pre-crisis period and firm leverage, exports or balance-sheet variables. The results for the crisis period are somewhat different. We now find evidence of a positive export-expansion effect (the export/sales ratio has a positive, statistically significant coefficient). However, the leverage ratio and the short-term debt ratio are not significantly related to sales growth (column 1). The financial variables continue to be statistically insignificant when they are interacted with firm size (column 2 and 3). In the first specification where financial variables are not interacted with firm size, the coefficient on the net foreign debt ratio is positive, although not statistically significant. The literature reports similar findings of no effect or a positive balance-sheet effect for publicly-listed firms.24 Foreign debt continues to be insignificant when it is interacted with firm size (column 2). These findings would suggest that there is no role for foreign debt and other financial variables in explaining firm performance. However, when the net foreign debt ratio is decomposed into short24 See Bleakley and Cowan (2008, 2009), Benavente et al. (2003), Bonomo et al. (2003), Forbes (2002), and Luengnaruemitchai (2003).

term and long-term debt (column 3), the coefficient on the short-term foreign debt ratio turns negative and the coefficient on the interaction term between short-term foreign debt and firm size becomes positive, even though they are only marginally significant at the 10% level. This implies that a pre-crisis net short-term foreign debt ratio is negatively associated with firm performance for small firms and is positively associated with performance for large firms. This is the first clue that short-term debt has a differential effect on the performance of large firms relative to small firms. The evidence is weak in the sample of publicly-listed firms because even “small” publicly-listed firms are fairly large. Next we turn to the full sample of firms where the small size effect will become clearer.

4.2.2. Results for the full sample Table 5 repeats the analysis for the full sample of firms that includes small, privately-held firms. Note that this sample is about four times the size of the sample in Table 4. This analysis will still miss the impact of firm exit, however, as we include only those firms that survive for the three-year interval (1994–1996 in the pre-crisis regression and 1996–1998 in the crisis regression). The results in the pre-crisis period are generally similar to those for the sample of publicly-listed firms except the effect of chaebol status becomes significantly positive in the full sample of firms (the right panel of Table 5). As in the sample of publicly-listed firms, firm age and size are negatively associated with sales growth performance, while firm size becomes insignificant when it is interacted with the other financial variables. Moreover, there is no evidence of an export effect, no significant effect for leverage, short-term debt or foreign debt

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

ratios prior to the crisis, similar to the results for publicly-listed firms in the pre-crisis period. Turning to the crisis regression results in the left panel of Table 5, we see that the chaebol dummy continues to have significant positive association with sales growth. Age is no longer significant. As in the precrisis regression, firm size is negatively associated with firm sales only when we do not allow the effects of the other variables to vary by firm size. Short-term debt exposure now has significant effects that vary with firm size. In the crisis, greater exposure to short-term debt is associated with faster sales growth rates for small firms, but slower sales growth rates for large firms. In contrast to the pre-crisis period, there is a robust relationship between exports and firm sales in crisis: the coefficient on export status is positive and strongly significant across all three specifications. The effect is also economically significant. A 10% increase in the pre-crisis export sales ratio is associated with an increase in sales growth of approximately 2% during the crisis. This export-expansion effect is similar to what we find in the sample of publicly-listed firms. The main difference across the two samples of firms is the balancesheet effect. In contrast to the findings for publicly-listed firms, the full sample shows strong evidence of a negative balance-sheet effect on small firms. If the financial variables are not interacted with size, the specification in column 1 continues to yield a puzzling positive coefficient on foreign debt. Different from the sample of publicly-listed firms, the coefficient is now statistically significant in the full sample. This seems to suggest that firms entering the crisis with higher foreign debt ratios had better performance during the crisis. When foreign debt is interacted with firm size (columns 2 and 3), however, the picture changes. The coefficient on the net foreign debt ratio in column 2 turns significantly negative and the coefficient on the interaction term

between foreign debt and size is significantly positive. The full sample shows strong evidence of a negative balance-sheet effect on small firms. Holding all the other variables constant, a 1% larger foreign debt ratio affects sales growth by (− 0.621 + 0.029 × size) percent, which monotonically increases with firm size. The impact is negative for small firms, but positive for large firms. The critical size below which the effect of foreign debt is negative is 21.5 in terms of log real assets and corresponds to a firm at about the 7th percentile in the size distribution. Thus, the negative balance-sheet effect appears only for very small firms. The negative balance-sheet effect on small firms is more prominent through short-term foreign debt. The coefficient on short-term foreign debt in column 3 is significantly negative (− 1.647). There is again an interaction effect with size—for large firms in the sample, the impact of short-term foreign debt is positive while for small firms the effect is negative. In this case, the cut-off point is 24.2 in terms of log real assets and corresponds to a firm at the 73rd percentile in the size distribution. Thus, for firms with assets below the 73rd percentile, an increase in the short-term foreign debt ratio is associated with a lower sales growth rate, all else equal. The effects are economically significant. Consider a firm with assets at the 10th percentile (log real asset of 21.7). A 10% increase in the short-term foreign debt ratio prior to the crisis is associated with a 1.7% lower rate of sales growth during the crisis. The findings of the balance-sheet effect are summarized in Fig. 5, which plots the marginal effect of the short-term foreign debt ratio on firm sales growth across firms with different size. The upper panel plots data for the full sample, and the lower panel for publicly-listed firms. The left panel shows the crisis period and the right panel shows the pre-crisis period. The solid line is the estimated marginal effect

Percentage Point

Full Sample, Crisis

Full Sample, Pre-crisis

1.5

1.5

1

1

0.5

0.5

0

0

-0.5

-0.5

-1

-1

-1.5 18

20

24 26 28 22 log(Real Assets)

30

-1.5 18

Percentage Point

Publicly-Listed Firms, Crisis 1.5

1

1

0.5

0.5

0

0

-0.5

-0.5

-1

-1

20

24 26 28 22 log(Real Assets)

30

20

22 24 26 28 log(Real Assets)

30

Publicly-Listed Firms, Pre-crisis

1.5

-1.5 18

219

-1.5 18

20

22 24 26 28 log(Real Assets)

30

Fig. 5. Marginal effect of short-term foreign debt on sales growth. Note: The solid line plots the estimated marginal effect of short-term foreign debt on sales growth for firms with different sizes. The dashed lines are the 90% confidence intervals. The upper panel is for the full sample, and the lower panel is for the publicly-listed firms. The left panel is for the crisis period, and the right panel is for the pre-crisis period. The marginal effects are computed using the regression estimates in columns 3 and 6 of Tables 4 and 5.

220

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

Table 6 Joint distribution of foreign debt and export status. Firms with foreign debt

Full sample Asset bins 1 2 3 4 Publicly-listed firms

Firms with no foreign debt

Total number

Fraction of exporters (%)

Fraction of non-exporters (%)

Total number

1528

28.80

71.20

2427

Fraction of exporters (%) 8.49

91.51

Fraction of non-exporters (%)

70 354 448 656

10.00 22.32 30.80 32.93

90.00 77.68 69.20 67.07

919 635 541 332

5.01 11.02 10.54 9.94

94.99 88.98 89.46 90.06

601

34.28

65.72

343

13.12

86.88

Note: A firm is classified as an exporter if its export/sales ratio is positive, and as a non-exporter otherwise. A firm is classified as a foreign debt holder if its foreign debt holdings are positive, and as a non-foreign-debt holder otherwise. The statistics are measured in 1996. The asset bins are based on real assets. Asset bin 1 is the smallest quartile and asset bin 4 is the largest quartile in the firm size distribution.

and the two dashed lines are the 90% confidence intervals. Clearly, the marginal effect of the short-term foreign debt ratio is not significantly different from zero during the pre-crisis period for either sample. During the crisis, we observe a size-dependent effect: significantly negative for small firms but positive for large firms for the full sample. The publicly-listed firms show a similar relation though not statistically significant. The exact interpretation of the positive balance-sheet effect for large firms is unclear. One possibility is that large firms are more likely to hedge exchange rate risk and are more likely to have access to other means of financing during the crisis. Ideally, a measure of the firm's net foreign currency denominated liabilities would reflect the firm's exposure on the liabilities side and its hedging activities on the asset side. Unfortunately, our dataset does not include information on foreign currency denominated assets other than cash holdings so there remains significant measurement error in our net foreign liabilities term. We also do not capture a firm's access to capital markets through equity markets or other off-balance-sheet sources of finance. Both of these factors are likely to be more important for large firms than for small firms, which gives us the confidence in the results for small firms, and also could explain the paradoxical results for large firms.25 The cross-section results based on the full sample support the view that both the export-expansion channel and the negative balance-sheet channel played a role during the crisis, with a particular role for exposure to short-term foreign debt. An interesting question is whether firms that were exposed to balance sheet risk were also exporters, and therefore were at least partially hedged from the negative impact of the exchange rate devaluation. Table 6 shows the decomposition of firms by export status and foreign debt holdings. The table shows that the share of non-exporters among firms that held foreign debt is 71% in the full sample and 66% in the publicly-listed sample. (The breakdown is similar for short-term foreign debt.) Thus, a significant fraction of firms that entered the crisis with foreign debt did not have a natural hedge for their currency exposure. The ratio of “non-hedged” to “hedged” firms—as measured by export status—is higher in the full sample than in the publicly-listed firms: 2.5 (= 71/29) versus 1.9 (= 66/ 34). We find that the ratio decreases with firm size; the ratio is above 9 (= 90/10) for the smallest quartile and about 2 (= 67/ 33) for the largest quartile, indicating that small firms with foreign debt holdings were more exposed to exchange rate risk.

25 The empirical literature documents that large, publicly listed firms tend to engage in multinational activity. See Horst (1972) and Grubaugh (1987) for the U.S. firms, and Jeon (1992) for Korean firms. Also, Geczy et al. (1997) and Allayannis and Ofek (2001) show that large firms are more likely to hedge foreign exchange risk using foreign currency derivatives.

4.3. Firm exit during the crisis The cross-section results pertain to firms that survived the crisis. We now perform an analysis of the factors that predict a firm's liquidation before and during the crisis. We run the following nonlinear probability regression on the panel of both surviving and exiting firms for the precrisis and crisis period:
PðEXITi ¼ 1Þ ¼ Φ α þ β FDi; −1 þ γ CHARi; −1 ;

ð3Þ

where P denotes the probability, EXIT is an indicator function of firm liquidation, and Φ denotes the logistic function. In the crisis period, the dependent variable is 1 if the firm exited in 1997 or 1998, and 0 otherwise. The independent variables are firm-specific observations in 1996, to capture the pre-crisis characteristics of the firm. In the pre-crisis period, the dependent variable is 1 if the firm exited in 1995 or 1996, and 0 otherwise. Firm characteristics on the right hand side are measured in 1994. Firm characteristics include chaebol status, age, one-digit industry dummy, size, export/sales ratios, leverage ratios, short-term debt ratios, and net (short-term and long-term) foreign debt ratios. If these firm characteristics do not fully capture aspects of the firm's quality that predict the firm's exit, the observed effects of these variables may be spurious. To address this issue, we include Altman's (1968) Z-score to control for the ex-ante risk of default.26 Comparing results before and during the crisis reveals whether the factors that are correlated with the likelihood of firm exit during the crisis are different from those before the crisis. The coefficients of the logit regressions are reported in Table 7. Turning first to the pre-crisis period (columns 4, 5 and 6) we see that relative to surviving firms, exiting firms tend to be younger, smaller and carry more debt in the year preceding liquidation. As expected, firms with higher Z-scores are significantly less likely to exit. Export status does not significantly affect the likelihood of exit. Foreign debt does not significantly predict firm exit in the pre-crisis period, independent of whether it is interacted with firm size or whether it is decomposed by maturity. Turning next to the crisis period (columns 1, 2 and 3), we see that chaebol status and firm size do not affect the likelihood of exit.27 Younger firms are more likely to exit. Leverage becomes a more significant predictor of exit in the crisis period. For a nonchaebol manufacturing 26 The Altman's Z-score is an index of the risk of default. Each firm is assigned a score according to the following formula in Altman (2000): Z = 0.717 T1 + 0.847 T2 + 3.107 T3 + 0.420 T4 + 0.998 T5, where T1 is (current assets-current liabilities)/total assets, T2 is retained earnings/total assets, T3 is earnings before interest and taxes/total assets, T4 is book value of equity/total liabilities, and T5 is sales/total assets. The higher the Z-score, the lower the probability of default. The Z-score is frequently included in survival/exit analyses to control for ex-ante risk of default. See Zingales (1998), for example. 27 Note that in the pre-crisis period, no chaebol firms were liquidated, and therefore we cannot compare across samples. Even during the crisis, chaebol firms tended to be restructured and absorbed by other firms rather than undergo complete liquidation.

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

221

Table 7 Coefficients of logit exit regressions. Crisis 1 Chaebol dummy Age Size Leverage ratio ST debt ratio Z-score

2

Pre-crisis 3

4

5

6

−0.571 (0.424) −0.132⁎⁎ (0.053)

−0.480 (0.426) −0.094⁎ (0.054)

−0.419 (0.425) −0.090 (0.055)

−0.382⁎⁎⁎ (0.083)

−0.360⁎⁎⁎ (0.086)

−0.350⁎⁎⁎ (0.088)

−0.004 (0.008) 5.409⁎⁎⁎

−0.004 (0.008) 5.364⁎⁎⁎

−0.003 (0.009) 5.359⁎⁎⁎

−0.031⁎⁎ (0.013) 2.648⁎⁎

−0.031⁎⁎ (0.013) 2.643⁎⁎

−0.032⁎⁎ (0.013) 2.591⁎⁎

(0.815) 1.112⁎⁎⁎ (0.362) −0.532⁎⁎⁎

(0.816) 1.127⁎⁎⁎ (0.363) −0.528⁎⁎⁎

(0.821) 0.958⁎⁎⁎ (0.365) −0.556⁎⁎⁎

(1.193) 0.672 (0.594) −0.842⁎⁎⁎

(1.194) 0.671 (0.595) −0.843⁎⁎⁎

(1.195) 0.689 (0.600) −0.841⁎⁎⁎

(0.087)

(0.088)

(0.090)

(0.188)

(0.189)

(0.190)

Export/sales ratio

0.134 (0.445)

0.143 (0.449)

0.195 (0.463)

0.808 (0.609)

0.830 (0.607)

0.853 (0.595)

Foreign debt ratio

0.082 (0.146)

6.805⁎⁎ (3.431) −0.287⁎⁎

−0.310 (0.328)

5.189 (3.679) −0.236 (0.163)

Size ∗ foreign debt ratio

(0.146) 27.63⁎⁎⁎ (10.680) −1.165⁎⁎ (0.460) 3.035 (3.105) −0.135 (0.133)

ST foreign debt ratio Size ∗ ST foreign debt ratio LT foreign debt ratio Size ∗ LT foreign debt ratio Observations Pseudo R-squared

4433 0.182

4433 0.185

4393 0.186

1.604 (7.271) −0.095 (0.338) 15.140 (10.110) −0.656 (0.439) 2943 0.201

2943 0.201

2936 0.202

Note: The dependent variable is either 1 if the firm exits in 1997 and 1998 or 0 otherwise for the crisis regressions, and is either 1 if the firm exits in 1995 and 1996 or 0 otherwise in the precrisis regressions. The independent variables are for 1996 in the crisis regressions and for 1994 for the pre-crisis regressions. Firms with negative net worth are excluded from the sample. Robust standard errors are reported in parentheses. All regressions include industry dummies at the one-digit level. ⁎⁎⁎ Denotes a p-value less than 1%. ⁎⁎ Denotes a p-value less than 5%. ⁎ Denotes a p-value less than 10%.

firm with characteristics at the mean level, the marginal effect of a higher leverage ratio on the exit probability is 3.7 times larger during the crisis than before the crisis. The coefficient on the short-term debt ratio turns highly significant in the crisis period, while it is not significant in the pre-crisis period. The Z-score continues to strongly predict firm exit during the crisis, though the absolute magnitude of the coefficient is smaller in the crisis. The export/sales ratio is not statistically significant, suggesting that exporting does not lower the likelihood of default in the crisis. The impact of foreign debt on firm exit during the crisis is dramatically different from the pre-crisis period. When foreign debt is interacted with firm size (columns 2), foreign debt has a statistically significant effect on firm exit even after controlling for the ex-ante probability of default. The effect of foreign debt on the exit probabilities varies with firm size, suggesting that small firms with foreign debt are more likely to exit while large firms are less likely to exit. This suggests that there is a negative balance-sheet effect not only for small, surviving firms but also for small firms at the exit margin. Similar to the crosssection analysis, the negative balance-sheet effect for small firms operates through short-term foreign debt. When foreign debt is decomposed by maturity (column 3), larger short-term foreign debt significantly predicts a higher exit probability for small firms, while long-term foreign debt does not significantly predict firm exit during the crisis. In a nonlinear model, the marginal effect of independent variables varies by observation and depends on all the covariates in the model. To examine the marginal effect of foreign debt across the two periods, Fig. 6 plots the marginal effect of short-term foreign debt on the exit probability (y-axis) before and during the crisis for nonchaebol

manufacturing firms with different size. We fix foreign debt ratios at the mean level conditional on having positive foreign debt and all the other variables at the mean level of the sample. The solid line is the estimated marginal effect and the two dashed lines are the 90% confidence intervals. The marginal effect of short-term foreign debt on the probability of exit is significantly positive for small firms and significantly negative for large firms during the crisis. In contrast, the marginal effect of short-term foreign debt pre-crisis is close to zero for most observations and not statistically significant. Thus, a larger short-term foreign debt ratio raises exit probabilities of small firms only during the crisis. During the crisis, for firms below (above) the 65th percentile of the size distribution, a larger short-term foreign debt ratio predicts a higher (lower) likelihood of exit. For example, for a firm with size at the 10th percentile and all other variables at the mean level, an increase in the pre-crisis net short-term foreign debt ratio of 10% is expected to increase the probability of exit during the crisis by 1.3%. In contrast, if the firm is in the top decile, a 10% increase in the short-term foreign debt ratio is expected to decrease the probability of exit by 0.4%. We now quantify the impact of foreign debt on sales growth by focusing on small firms that are negatively impacted by the balancesheet effect.28 Consider a counterfactual experiment in which these small firms enter the crisis with larger pre-crisis short-term foreign debt ratios by one standard deviation (12 percentage points). Our

28 As shown in Section 4.2, firms with size less than the 73rd percentile are negatively impacted by short-term foreign debt holdings during the exchange rate depreciation. In 1996, the value added share and the sales share of these firms in the aggregate economy are 10.4% and 10.6%, respectively. The share of employment accounted by those small firms is much larger at 49.8%.

222

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

Percentage Point

Crisis

Pre-crisis

0.8

0.8

0.6

0.6

0.4

0.4

0.2

0.2

0

0

-0.2 18

20

22 24 26 28 log(Real Assets)

30

-0.2 18

20

22 24 26 28 log(Real Assets)

30

Fig. 6. Marginal effect of short-term foreign debt on the probability of exit. Note: The solid line plots the estimated marginal effect of short-term foreign debt on the exit probability for nonchaebol manufacturing firms with different sizes, the short-term and longterm foreign debt ratios at the mean levels conditional on having positive foreign debt, and all the other variables at the mean level of the sample. The dashed lines are the 90% confidence intervals. The left panel is for the crisis period, and the right panel is for the pre-crisis period. The marginal effects are computed using the regression estimates in columns 3 and 6 of Table 7.

logit regression results predict that the exit rate of firms in the smallest quartile rises by 1.1 percentage points in the counterfactual. Our survival analysis suggests that firms in the smallest quartile see a decline in the sales growth rate by 2 percentage points. Incorporating both effects, we find that increasing short-term foreign debt ratios by one standard deviation is associated with a 3-percentage-point decline in sales growth of firms in the smallest quartile. The impact on the economy is small in contrast. The overall exit rate rises by 3 basis points, the aggregate survivor sales growth rate falls by 10 basis points, and both margins imply a decline of aggregate sales growth by 12 basis points. This result is not surprising given our finding of the balance-sheet effect on only small firms. A key insight from our analysis is that the impact of foreign debt depends critically on what types of firms take on foreign debt. If foreign debt is concentrated in the balance sheets of small firms, the decline in their sales growth may be large. Further, the exit margin explains the majority of the decline of sales growth for small firms in the sample. For example, the exit margin accounts for 64% of the decline in sales for the smallest quartile, while it accounts for 38% of the overall sales growth decline. These numbers underscore the importance of taking firm exit into account when evaluating the effects of the crisis. In order to illustrate the natural hedging roles played by exports during the crisis, consider another counterfactual experiment in which we set all exporters' export/sales ratio to zero. The aggregate survivor sales growth rate declines by 2 percentage points, and the exit rate rises by 3.5 percentage points. Both margins together lead to a decline of the overall sales growth by 5 percentage points, which is not surprising given that exporters are generally large. This experiment illustrates the substantial benefit from hedging through exports during an exchange rate crisis, though this benefit is likely unavailable for small firms, who are more likely to be non-exporters. 5. Robustness checks In this section, we conduct a series of checks to confirm the robustness of the negative balance-sheet effect on small firms and the positive export-expansion effect to alternative specifications. Tables of results are delegated to Appendix 2. 5.1. Robustness to alternative performance measures We start by examining whether our results remain intact when alternative firm performance variables are used. An increase in the domestic currency value of foreign debt following an unexpected

exchange rate depreciation works as an exogenous shock to a firm's net worth, and thus leads to an increase in financing costs. To the extent that firms' production inputs have to be financed, the balance-sheet effects will reduce purchases of inputs to production. We use the ratio of investment to the capital stock at the end of the previous year to measure capital input,29 the growth rate of employment and the growth rate of personnel and selling costs to measure labor input, and the growth rate of raw material costs to measure intermediate inputs. We find that during the crisis foreign debt holdings negatively affect investment by small firms. When we regress investment in 1999 on the precrisis firm characteristics to control for planning delays, long-term foreign debt is detrimental to investment by small firms. Though employment shows no significant balance-sheet effect,30 the alternative measure of labor input—personnel and selling costs—and the raw material costs show strong evidence of the negative balance-sheet effect on small firms, especially through short-term foreign debt.

5.2. Robustness to export responses during the crisis Given the significant export-expansion effect identified in the analysis, firms might respond to the exchange rate depreciation by increasing exports during the crisis. This ex-post hedging channel through exporting might impact our balance-sheet results, especially for large and publicly-listed firms. In our dataset, we find that about 5% of precrisis non-exporting firms begin exporting during the crisis and about 10% of pre-crisis exporting firms increase their export sales ratios during the crisis. The median export/sales ratio of these new exporters is 11%, and the median increase of the export/sales ratios of existing exporters

29 Investment is constructed based on the book values of tangible assets from the balance sheet. This measurement is accurate if book value measures nominal value at historical cost. In the 1990s, the Korean accounting standards required that tangible assets should be reported at historical costs, and asset revaluation was allowed when the PPI rose more than 25%. According to the data released from National Tax Service, less than 1% of the total firms revalued their assets in the 1990s. Also, a large fraction of asset revaluation was on non-depreciable assets (land, stocks, etc.) which do not constitute physical investment. Alternatively, investment could be constructed based on the information from the cash flow statement which directly reports the costs of purchasing tangible assets. The Korean law started to require that firms report the cash flow statement starting in 1994. We found that the quality of the cash flow statement was not very good in the 1990s. Thus, we use the balance sheet information to construct investment measure. See the appendix for a more detailed description of the calculation of investment and capital stock. 30 Employment is generally considered stable and long-term in South Korea due to cultural aspects and the influence of labor unions. Also, employment is poorly measured in financial statements, and only about half of the sample firms report employment data. Both factors might prevent us from finding meaningful results.

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

is 3.8%. As expected, large firms are more likely to start exporting or to increase their export intensity. We conduct the robustness check in two ways. In the first approach, we exclude firms that changed their export behavior at the extensive and/or intensive margin during the crisis and re-run the baseline regressions. In the second approach, we add a dummy variable for these firms in the baseline regressions. We find that the coefficients of the variables of interest (say the export/sales ratio, the foreign debt ratio, and the interaction terms between the foreign debt ratio and size) barely change across these experiments. The dummies on firms that change export status are not statistically significant. Our interpretation is that since the observed changes in the exporting status and the exporting intensity in the crisis are modest, the effects of this ex-post hedging channel in the crisis on firm performance are limited. 5.3. Robustness to fixed-effect panel analysis We repeated the estimation using fixed-effect panel regressions, which potentially would ameliorate the concern about omitted variables. It is worth emphasizing that one of our major reservations about this analysis is that the panel is very short; the fixed-effect regression estimates coefficients of two time periods (pre-crisis versus crisis) with only five years of data. The panel results are consistent with our baseline results. We observe one slight difference relative to the cross-section results when considering shortterm and long-term foreign debt. The statistical significance of short-term debt shows up in the investment regression instead of in the sales regression. 5.4. Robustness to a balanced sample As noted in the description of the data in Section 3, the number of firms in the sample increases substantially over time due to the fast growth rates of Korean firms and the fixed nominal cutoff that determines inclusion (by KIS) in the database. Our baseline analysis includes firms that are present in 1994–96 for the pre-crisis regressions and firms that are present in 1996–98 for the crisis regressions. By doing so, we include more firms in the crisis sample than in pre-crisis sample. We repeated the analysis for firms that are present in all years between 1994 and 1998. The baseline results are robust to holding the samples balanced across the two periods, with one exception that the coefficients of long-term foreign debt, instead of those of short-term foreign debt, are significantly negative for small firms. The new balanced sample has 33% fewer firms than the baseline sample in the crisis period, and excludes disproportionately small firms, which changes the results to some extent. 5.5. Robustness to the import channel and foreign ownership Exchange rate depreciations tend to adversely affect firms relying on imported intermediate inputs. One potential concern is that if importers were also more likely to have foreign debt, the finding that firms with larger foreign debt exposure have worse performance in the crisis might be due to this import channel, rather than the balance-sheet channel. The ideal way to address this concern would be to control for firm imported input demand in the analysis. Unfortunately our firmlevel dataset does not have information on imported intermediates. We instead explore the cross-industry variation in the imported intermediate input demand. We first compute the imported input coefficient (IIC)—the share of imported inputs in total intermediate inputs—at the two-digit industry level using Korean input–output tables. The IICs have substantial variations across two-digit industries: from almost zero in the non-metallic minerals sector and 0.65 in the coal product sector. We then run the baseline regressions industry-by-industry to obtain the coefficients of foreign debt on sales growth for a firm at the bottom 10th percentile of the size distribution in that industry. Finally,

223

we examine whether the coefficients are correlated with the IICs across industries. If the import channel accounts for our findings, we should see a negative relationship between the estimated coefficients on foreign debt and industry-level IICs. That is, an industry with a high reliance on imported intermediate inputs would appear to be more negatively affected by the foreign debt ratio. We find no systematic relation between foreign debt coefficients and IICs, which suggests that the import channel is not the force behind our key findings. Kalemli-Ozcan et al. (2010) document that in the case of Latin American countries, foreign-owned firms (affiliates) suffer less during banking crises because they have better access to liquidity through the associated parent companies. In order to control for access to liquidity, which might co-vary with foreign debt holdings and firm performance, we include a dummy for foreign ownership. While the results are not reported to conserve space, the foreign ownership dummy turns out to be insignificant and the inclusion of it leads to negligible changes in the regression results. 6. Conclusion Using Korean firm-level data on both publicly-listed and privatelyheld firms and firm exit data, this paper finds evidence of a balance-sheet effect and an export-expansion effect. Before the crisis, firm sales growth was uncorrelated with foreign debt holdings and export sales. During the crisis, however, small firms entering the crisis with higher levels of foreign debt, and in particular, short-term foreign debt, experienced larger declines in sales growth. Firms with higher export sales ratios experienced smaller declines in sales growth during the crisis. In addition, we find that small firms with large holdings of short-term foreign debt are significantly more likely to go bankrupt during the crisis than firms with less short-term foreign debt on their balance sheet. The exit margin accounts for a large fraction of small firms' adjustment during the crisis. There are two caveats to these conclusions. The first is that the results in this paper pertain primarily to differential firm performance in the cross-section. As shown in Fig. 4, most of the variation in the data is at the macro level. That is, our results can only explain whether firms with more foreign debt holdings have sharper declines in sales than firms with smaller holdings and we do not claim to provide an explanation for the overall decline in firm sales. Second, the regression analyses take firm characteristics (size, debt ratios, export status, etc.) as given in explaining next period's sales growth. Obviously, many firm characteristics are themselves choice variables, and a complete model would endogenize the full menu of firm characteristics, including firm debt, exposure to foreign currency risk and export status. We leave a more complete analysis that would address these caveats for future research. One might ask, what is the relevance of the Korean crisis for understanding the risks of foreign currency borrowing today? After all, the crisis occurred nearly 20 years ago, and surely there are greater opportunities for hedging exchange rate risk now relative to the late 1990s. One interpretation of our results is that we provide a careful autopsy of a historically important, but by-gone era, and that the risks of foreign currency exposure are well behind us. A quick perusal of the financial pages suggests that this is probably a mistaken view. Negative fallout from sudden capital flight from emerging markets continues to be a global concern. Many countries have built up buffer stocks of reserves in an attempt to insure against sudden changes in the capital account and the exchange rate, but it is unlikely any country is fully protected. Our findings suggest that in assessing the risk of exchange rate exposure, it is important to know precisely who is carrying that risk. In the case of Korea, small, nonexporting firms found themselves bearing the brunt of the economic adjustment.

224

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

Appendix 1. A simple model of the balance-sheet effects We develop a one-period model of a firm operating under credit constraints to illustrate the connection between exchange rates, foreign debt, export share, exit and firm performance. The model is quite simple, but it is intended to provide a framework for the empirical analysis that follows and to clarify the assumptions we make in approaching the data. Consider the situation of a firm at the time of the crisis in 1997. The crisis is characterized by a large depreciation in the Korean won, and the percentage increase in the exchange rate is denoted by Δewon/$ N 0. Prior to the crisis, the firm has an initial net worth of n0 and a net foreign debt over net worth ratio of d0. Assume that market participants (firms as borrowers and banks as lenders) failed to anticipate the crisis and therefore failed to take measures to insure against exchange rate risk. The net worth after the exchange rate shock is n1 = n0 − d0n0Δe.31 The larger the share of foreign debt in firm net worth d0, the bigger the impact of the change in the exchange rate on the firm's net worth, all else equal. Looking forward, the firm faces uncertainty over aggregate risk, denoted by A, with mean Ā. The aggregate risk captures the negative macroeconomic impact of the crisis affecting the entire distribution of firms. The firm also faces an idiosyncratic productivity shock z. The firm-specific shocks are assumed to be i.i.d. with a mean of z. Firms differ in their ex post realization of the idiosyncratic shock to productivity, as well as in the ex ante mean level of productivity. Firms also differ in the extent to which they export their final good. We assume that the share of exports in total sales, ϕ, is exogenous. α γ β The firm's output is y1 = A1z1(kα1 l1− ) M1, where α denotes the share of capital in value added, γ denotes the value added share in output, and β 1 denotes the intermediate input share. Assume that γ + β b 1. We denote the aggregate and idiosyncratic productivity shock realization with A1 and z1, respectively. k1, l1 and M1 denote the capital, labor and intermediate inputs. Given the decreasing returns to scale specification, firm size is determined and is increasing in z. Following Bernanke et al. (1999), we assume that firms decide on capital, labor and intermediate inputs prior to the realization of aggregate and idiosyncratic shocks. The working capital needed to finance the operation of the firm may be covered by the firm's net worth or with a bank loan, b1: P 1 M 1 þ R1 k1 þ W 1 l1 ¼ n1 þ b1 ; where P1 denotes the price of intermediate input, R1 denotes the cost of capital, and W1 denotes the wage rate. For simplicity, we assume that the firm's need for working capital exceeds its net worth so it must borrow from banks in order to produce.32 Firms may borrow from banks at the rate r 1 ðn1 ; ϕ; z; AÞ ¼ r f ð1 þ ηðn1 ; ϕ; z; AÞÞ where rf denotes the world risk free rate, and ηðn1 ; ϕ; z; AÞ denotes the financing premium. The premium is conditional on the realization of aggregate risk and on the expectation of idiosyncratic risk. As is standard in models with costly state verification, the aggregate shock is freely verifiable, but the idiosyncratic shock is costly to verify. Furthermore, the risk premium is decreasing in the firm's net worth, its expected idiosyncratic shock and the aggregate shock. The financing premium is also decreasing in the export share because with imperfect pass-through exports increase firm sales and decrease the ex post likelihood of costly verification. This will be clear when we discuss the firm's problem below. In the immediate aftermath of the exchange rate crisis, the problem facing the firm is     γ   α 1−α V n1 ; ϕ; z; A ¼ max 1−ϕ þ ϕ 1 þ Δewon=$ ð1−π Þ Az k1 l1 Mβ1 −EA ½r 1 ðn1 ; ϕ; z; AÞðW 1 l1 þ R1 k1 þ P 1 M1 −n1 Þ: k1 ;l1 ;M1

ð4Þ

The firm chooses factor and intermediate inputs to maximize the value of the firm, endogenizing the cost of financing working capital and conditioning on its expected level of productivity. Note that the value of firm output is measured in domestic currency units, and π is an indicator of exchange rate pass through. If pass-through is perfect π = 1, the won price of Korean exports falls by the same amount of the won depreciation. In this case the exchange rate change has no effect on the value of exporters' output measured in Korean won. If pass-through is imperfect π b 1, there is a benefit to being an exporter as the value of output is increasing in the export share. It is well known empirically that the exchange rate pass through is imperfect. Firm sales and firm value V both increase with the export intensity ϕ during an exchange rate depreciation. Exports are therefore a hedge to an exchange rate depreciation. The firm's problem illustrates the role of the balance-sheet effects and export expansion effects for firm performance following an exchange rate shock. Consider first the balance-sheet effect. Depreciation of the exchange rate reduces the firm's net worth n1 if the firm has foreign-currency denominated debt. A reduction in net worth tightens the working capital constraint; all else equal, the firm requires more bank finance to attain the same level of output. The reduction in net worth also increases the cost of bank finance r1. Consequently, both input demands and sales decline, leading to worsening firm performance. Turning to the export expansion effect, we see that exports affect the firm's decision in two ways. First, as long as there is less than complete pass through, the value of output and input demand are increasing in the share of export sales ϕ. This is the classical export expansion effect resulting from an exchange rate depreciation. Second, as a consequence of the balance sheet channel, the interest rate premium r1 decreases with export intensity ϕ. Thus, firms with a larger share of exporting sales contract less during the crisis. 31 Details of a firm's balance sheet and the timing of events are as follows. At the end of date 0, a firm has domestic asset A0, domestic debt D0, foreign asset A⁎0e0 and foreign debt D⁎0e0, where e0 denotes the won/dollar exchange rate, A⁎0 and D⁎0 are in unit of dollar, and A0 and D0 are in unit of won. The net worth is n0 = A0 − D0 + (A⁎0 − D⁎0)e0 = N0 + N⁎0, where N0 denotes the domestic net worth, and N⁎0 denotes the foreign net worth in unit of won. Note that the foreign net worth is also the opposite of the net foreign debt. Now let's define d0 as the share of net foreign debt in the total net worth, i.e., N⁎0 = − d0n0. Thus we can rewrite n0 = (1 + d0)n0 − d0n0. At the beginning of date 1, the unexpected exchange rate shock realizes: the won/dollar exchange rate rises (depreciates) by Δe ¼ ee10 −1 percent. Thus the net worth after the exchange rate shock is n1 = (1 + d0)n0 − d0n0(1 + Δe) = n0 − d0n0Δe. Then, the firm makes its investment, employment, intermediate input, production and sales decisions with the expectations of the aggregate shocks and firm-specific idiosyncratic shocks. 32 There is no need to differentiate the currency denomination of loans in this one-period model: arbitrage equates the interest rates of domestic currency and foreign currency denominated loans, so ex ante firms are indifferent between the two sources of financing.

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

225

Eq. (4) also illustrates the exit margin. If the continuation value of the firm is negative, i.e., costs of financing exceed the expected value of output   V n1 ; ϕ; z; A b 0; and the firm will exit. The balance-sheet and export-expansion effects discussed above have similar effects on the firm's decision to exit. Firms with a larger exposure to foreign debt will experience a larger decline in net worth, and they are more likely to exit. Exporting firms are less likely to exit. Smaller firms (due to lower expected productivity levels z) are more likely to exit. Lastly, the exit rate increases for all firms when the expected aggregate shock Ā is low. Firms with positive continuation value (who therefore opt not to exit) choose their demand for inputs and their level of bank loans before the realization of the idiosyncratic productivity and aggregate shocks. The finance premium drives a wedge between the marginal product of each input and its cost. However, the financial friction does not distort the relative usage of inputs. Therefore, as the cost of finance increases—say because of a negative shock to net worth—the firm reduces its demand for all production inputs, and output contracts. In sum, the model has three key predictions with respect to the balance-sheet effects. In response to an unexpected currency crisis, all else equal, firms with larger fractions of foreign currency denominated debt (i) will experience larger declines in net worth; (ii) conditional on survival, will contract more and have worse performance; (iii) and will exit with higher likelihood. We test these model predictions with Korean firm level data in Section 4. Appendix 2. Construction of the investment rate We follow Bayraktar et al. (2005) in constructing the investment rate. The investment rate is the ratio of real investment to the lagged replacement value of real capital stock. For real investment (It), nominal investment (Int) is first constructed by Int = Kbt − Kbt − 1 + Dept, where the book value of capital, Kbt, is calculated by subtracting land and lease assets from tangible assets (all in book values from the balance sheets), and depreciation, Dept, is taken from the cash flow statements. Real investment is nominal investment deflated by the capital goods price index. The replacement value of real capital stock (Kt) is calculated by iterating Kt = (1 − d) Kt − 1 + It backward, where It is real investment

Table A1 Cross-section regressions for alternative performance measures in crisis. Dependent variables

I/Klead −1

I/K−1

Employment growth

1

2

3

4

0.031 (0.030) 0.000 (0.001) −0.020 (0.024)

0.032 (0.030) 0.000 (0.001) −0.021 (0.024)

0.122⁎⁎⁎ (0.034) −0.001 (0.001) −0.086⁎⁎

0.119⁎⁎⁎ (0.034) −0.001 (0.001) −0.092⁎⁎

(0.040)

−0.294 (0.737) 0.014 (0.031) 1.040 (1.165) −0.044 (0.049)

−0.300 (0.743) 0.014 (0.031) 1.001 (1.200) −0.042 (0.050)

−0.110 (1.341) 0.002 (0.056) 0.455 (1.367) −0.022 (0.057)

Export/sales ratio

0.072⁎ (0.040)

0.074⁎ (0.041)

−0.062 (0.044)

Foreign debt ratio

−0.424⁎ (0.218) 0.018⁎⁎

−0.886⁎⁎ (0.357) 0.035⁎⁎

(0.009)

(0.015)

Chaebol dummy Age Size Leverage ratio Size ∗ leverage ratio ST debt ratio Size ∗ ST debt ratio

Size ∗ foreign debt ratio

−0.179 (0.915) 0.008 (0.036) −0.526 (0.354) 0.022 (0.015)

ST foreign debt ratio Size ∗ ST foreign debt ratio LT foreign debt ratio Size ∗ LT foreign debt ratio Observations R-squared

3959 0.038

3926 0.038

5

Personnel cost growth

9

10

(0.040)

0.010 (0.050) −0.001 (0.001) −0.100 (0.067)

0.027 (0.028) 0.000 (0.001) 0.004 (0.022)

0.015 (0.027) 0.000 (0.001) 0.003 (0.022)

0.064⁎⁎ (0.028) 0.001 (0.001) −0.017 (0.036)

0.060⁎⁎ (0.028) 0.001 (0.001) −0.019 (0.036)

−0.286 (1.344) 0.010 (0.056) 0.428 (1.404) −0.021 (0.059)

−3.370 (2.233) 0.140 (0.093) 0.708 (1.288) −0.029 (0.055)

−3.561 (2.244) 0.148 (0.094) 1.107 (1.315) −0.048 (0.056)

1.093 (0.680) −0.049⁎ (0.029) −1.503⁎

1.120 (0.685) −0.051⁎ (0.030) −1.534⁎

0.184 (1.151) −0.008 (0.048) 2.869⁎⁎

0.211 (1.149) −0.009 (0.048) 2.787⁎⁎

(1.380) −0.121⁎⁎ (0.058)

(1.412) −0.117⁎⁎ (0.060)

−0.048 (0.043)

−0.179⁎⁎⁎ (0.045)

−0.165⁎⁎⁎ (0.045)

8

(0.842) 0.0668⁎ (0.036)

(0.846) 0.0685⁎ (0.036)

0.086⁎⁎⁎ (0.033)

0.088⁎⁎⁎ (0.034)

−0.746⁎⁎⁎ (0.252) 0.033⁎⁎⁎

0.214 (0.341) −0.010 (0.014)

2313 0.047

2290 0.047

0.142⁎⁎⁎ (0.032)

(0.013) −2.426⁎⁎⁎

−1.893⁎⁎⁎ (0.674) 0.080⁎⁎⁎

(0.926) 0.101⁎⁎⁎ (0.038) −0.646 (0.508) 0.030 (0.021) 3979 0.064

0.148⁎⁎⁎ (0.033)

−0.910⁎⁎⁎ (0.312) 0.041⁎⁎⁎

(0.010) 0.982 (1.534) −0.035 (0.060) 0.820 (0.576) −0.037 (0.024)

(0.497) 0.048⁎⁎ (0.021) 3870 0.079

7

Raw material cost growth

0.030 (0.052) −0.001 (0.001) −0.099 (0.067)

−1.612 (1.137) 0.065 (0.045) −1.223⁎⁎

3904 0.080

6

3946 0.067

(0.027) −0.990⁎ (0.564) 0.046⁎⁎ (0.023) 3889 0.104

3854 0.104

Note: The dependent variables are indicated in the top row. I/K−1 denotes firm real investment in 1998 as a share of the replacement value of real capital stock in 1997. I/Klead −1 denotes firm real investment in 1999 as a share of the replacement value of real capital stock in 1998. The rates of employment growth, personnel cost growth, and raw material cost growth are measured as the real growth rates between 1997 and 1998. The independent variables are for year 1996 in all regressions. Firms with negative net worth are excluded from the sample. Robust standard errors are reported in parentheses. All regressions include industry dummies at the two-digit level and the lagged dependent variable. For publicly-listed firms, the results are broadly consistent with those of sales growth, and omitted to conserve the space. ⁎⁎⁎ Denotes a p-value less than 1%. ⁎⁎ Denotes a p-value less than 5%. ⁎ Denotes a p-value less than 10%.

226

Table A2 Ex-post hedging with exporting: full sample. Baseline

Extensive margin (new exporters) Excluded

Chaebol dummy Age Size

Size ∗ leverage ST debt ratio Size ∗ ST debt ratio Export/sales Foreign debt ratio Size ∗ foreign debt

0.418 (0.937) −0.016 (0.040) 2.492⁎⁎

0.501 (0.933) −0.020 (0.039) 2.316⁎⁎

0.750 (0.915) −0.031 (0.039) 2.442⁎⁎

0.835 (0.913) −0.035 (0.039) 2.246⁎⁎

0.520 (0.985) −0.021 (0.042) 2.422⁎⁎

0.608 (0.982) −0.025 (0.042) 2.196⁎⁎

0.751 (0.915) −0.031 (0.039) 2.428⁎⁎

0.834 (0.912) −0.035 (0.039) 2.236⁎⁎

0.135 (1.011) −0.004 (0.043) 2.511⁎⁎

0.216 (1.006) −0.008 (0.043) 2.313⁎⁎

0.749 (0.915) −0.031 (0.039) 2.439⁎⁎

0.833 (0.912) −0.035 (0.039) 2.245⁎⁎

(1.027) −0.103⁎⁎ (0.044)

(1.041) −0.094⁎⁎ (0.044)

(1.058) −0.106⁎⁎ (0.045)

(1.070) −0.098⁎⁎ (0.046)

(1.027) −0.104⁎⁎ (0.044)

(1.042) −0.094⁎⁎ (0.044)

(1.091) −0.103⁎⁎ (0.047)

(1.105) −0.093⁎⁎ (0.047)

(1.026) −0.103⁎⁎ (0.044)

(1.041) −0.094⁎⁎ (0.044)

(1.128) −0.107⁎⁎ (0.048)

(1.139) −0.098⁎⁎ (0.049)

(1.026) −0.103⁎⁎ (0.044)

(1.041) −0.094⁎⁎ (0.044)

−0.621⁎⁎ (0.311) 0.029⁎⁎ (0.012)

ST foreign debt ratio Size ∗ ST foreign debt LT foreign debt ratio Size ∗ LT foreign debt

−0.627⁎ (0.338) 0.029⁎⁎ (0.013)

3955 0.121

(0.030) 0.000 (0.001) −0.013 (0.029)

(0.028) 0.001 (0.001) −0.003 (0.027)

(0.028) 0.001 (0.001) −0.004 (0.027)

0.186⁎⁎⁎ (0.034)

−0.621⁎⁎ (0.311) 0.029⁎⁎ (0.012)

0.145⁎⁎⁎ (0.047)

0.150⁎⁎⁎ (0.047)

0.073⁎⁎⁎

0.180⁎⁎⁎ (0.037)

0.067⁎⁎

0.187⁎⁎⁎ (0.038)

−0.619⁎⁎ (0.311) 0.029⁎⁎ (0.012)

−0.502 (0.350) 0.024⁎ (0.014)

0.058⁎⁎ (0.028) 0.000 (0.001) −0.023 (0.031)

0.049⁎ (0.028) 0.000 (0.001) −0.024 (0.030)

(0.028) 0.001 (0.001) −0.004 (0.027)

0.066⁎⁎ (0.028) 0.001 (0.001) −0.005 (0.027)

0.156⁎⁎⁎ (0.047)

0.163⁎⁎⁎ (0.047)

0.072⁎⁎⁎

0.163⁎⁎⁎ (0.036)

0.170⁎⁎⁎ (0.037)

−0.625⁎⁎ (0.311) 0.029⁎⁎ (0.012)

−0.495 (0.385) 0.024 (0.015)

−1.647⁎⁎ (0.642) 0.068⁎⁎⁎

−1.912⁎⁎⁎ (0.677) 0.079⁎⁎⁎

−1.619⁎⁎ (0.645) 0.067⁎⁎⁎

−1.424⁎⁎ (0.693) 0.059⁎⁎

−1.652⁎⁎ (0.644) 0.068⁎⁎⁎

−1.771⁎⁎ (0.743) 0.074⁎⁎

−1.630⁎⁎ (0.644) 0.067⁎⁎⁎

(0.025) −0.864 (0.551) 0.042⁎ (0.022)

(0.027) −0.799 (0.560) 0.039⁎ (0.022)

(0.025) −0.863 (0.551) 0.042⁎ (0.022)

(0.027) −0.881 (0.609) 0.043⁎ (0.024)

(0.025) −0.859 (0.553) 0.042⁎ (0.022)

(0.029) −0.803 (0.621) 0.040⁎ (0.024)

(0.025) −0.871 (0.552) 0.042⁎ (0.022)

Export status change dummy Observations R-squared

(0.028) 0.001 (0.001) −0.005 (0.027)

0.071⁎⁎ (0.030) 0.000 (0.001) −0.012 (0.030)

Dummy

0.834 (0.912) −0.035 (0.039) 2.239⁎⁎

0.179⁎⁎⁎ (0.034)

0.063⁎⁎

Excluded

0.750 (0.915) −0.031 (0.039) 2.431⁎⁎

0.193⁎⁎⁎ (0.033)

0.067⁎⁎

Dummy

(0.028) 0.001 (0.001) −0.005 (0.027)

0.186⁎⁎⁎ (0.033)

(0.026) 0.001 (0.001) −0.014 (0.028)

0.073⁎⁎⁎ (0.028) 0.001 (0.001) −0.004 (0.027)

Excluded

(0.028) 0.001 (0.001) −0.004 (0.027)

0.179⁎⁎⁎ (0.034)

0.055⁎⁎

Dummy

0.062⁎⁎ (0.026) 0.001 (0.001) −0.013 (0.028)

0.173⁎⁎⁎ (0.034)

0.067⁎⁎

Intensive and extensive margins (firms who increased export/sales)

3921 0.125

3756 0.116

3723 0.120

0.045 (0.031) 3955 0.122

0.045 (0.032) 3921 0.126

3560 0.121

3535 0.125

−0.009 (0.019) 3955 0.122

−0.009 (0.020) 3921 0.125

3361 0.114

3337 0.118

0.015 (0.018) 3955 0.122

0.015 (0.019) 3921 0.126

Note: The dependent variable is firm sales growth between 1997 and 1998 and the independent variables are for year 1996. Firms with negative net worth are excluded from the sample. Robust standard errors are reported in parentheses. All regressions include industry dummies at the two-digit level and the lagged sales growth rate. ⁎⁎⁎ Denotes a p-value less than 1%. ⁎⁎ Denotes a p-value less than 5%. ⁎ Denotes a p-value less than 10%.

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

Leverage

0.072⁎⁎⁎

Intensive margin (existing exporters who increased export/sales)

Table A3 Ex-post hedging with exporting: publicly-listed firms. Baseline

Extensive margin (new exporters) Excluded

Chaebol dummy Age Size

Size ∗ leverage ST debt ratio Size ∗ ST debt ratio Export/sales Foreign debt ratio Size ∗ foreign debt

Excluded

Dummy

Excluded

Dummy

0.058 (0.036) 0.001 (0.001) 0.002 (0.057)

0.047 (0.036) 0.001 (0.001) 0.000 (0.057)

0.056 (0.037) 0.001 (0.001) −0.012 (0.058)

0.043 (0.037) 0.001 (0.001) −0.014 (0.057)

0.058 (0.036) 0.001 (0.001) 0.000 (0.057)

0.048 (0.036) 0.001 (0.001) −0.002 (0.056)

0.067 (0.049) 0.000 (0.001) −0.030 (0.072)

0.048 (0.049) 0.001 (0.001) −0.036 (0.072)

0.058 (0.036) 0.001 (0.001) 0.002 (0.057)

0.047 (0.036) 0.001 (0.001) 0.000 (0.057)

0.066 (0.051) 0.001 (0.001) −0.048 (0.075)

0.040 (0.053) 0.001 (0.001) −0.054 (0.074)

0.057 (0.036) 0.001 (0.001) 0.000 (0.057)

0.047 (0.036) 0.001 (0.001) −0.002 (0.057)

2.538 (1.704) −0.102 (0.068) −1.888 (1.895) 0.067 (0.073)

2.379 (1.664) −0.100 (0.066) −1.541 (1.810) 0.061 (0.072)

2.209 (1.720) −0.088 (0.069) −2.726 (1.963) 0.099 (0.075)

2.045 (1.686) −0.086 (0.067) −2.291 (1.859) 0.091 (0.074)

2.464 (1.697) −0.099 (0.068) −1.953 (1.889) 0.069 (0.073)

2.311 (1.656) −0.097 (0.066) −1.604 (1.804) 0.063 (0.072)

1.516 (2.124) −0.062 (0.085) −2.054 (2.134) 0.073 (0.082)

1.190 (2.029) −0.054 (0.082) −1.665 (2.010) 0.066 (0.081)

2.546 (1.710) −0.102 (0.068) −1.901 (1.901) 0.067 (0.073)

2.380 (1.667) −0.100 (0.067) −1.542 (1.810) 0.061 (0.072)

1.094 (2.174) −0.044 (0.087) −2.964 (2.223) 0.109 (0.085)

0.725 (2.083) −0.035 (0.084) −2.453 (2.082) 0.098 (0.083)

2.496 (1.705) −0.100 (0.068) −1.880 (1.898) 0.066 (0.073)

2.337 (1.661) −0.098 (0.066) −1.546 (1.808) 0.061 (0.072)

0.147 (0.091)

0.089 (0.102)

0.160⁎ (0.092)

0.096 (0.104)

0.218⁎⁎⁎ (0.061)

0.183⁎⁎⁎ (0.066)

0.511 (1.748) −0.010 (0.065)

0.234⁎⁎⁎ (0.062)

Size ∗ ST foreign debt LT foreign debt ratio Size ∗ LT foreign debt

0.195⁎⁎⁎ (0.068)

0.753 (1.848) −0.019 (0.069) −3.307⁎ (1.959) 0.129⁎ (0.074) 1.667 (3.211) −0.046 (0.120)

ST foreign debt ratio

944 0.191

934 0.224

0.227⁎⁎⁎ (0.061)

872 0.227

1.025 (2.013) −0.029 (0.075) −3.281⁎ (1.965) 0.128⁎ (0.075) 1.687 (3.210) −0.047 (0.119)

−3.291 (2.032) 0.129⁎ (0.077) 1.861 (3.245) −0.053 (0.121)

882 0.190

0.192⁎⁎⁎ (0.067)

0.536 (1.747) −0.011 (0.065)

Export status change dummy Observations R-square

Dummy

Intensive and extensive margins (firms who increased export/sales)

0.059 (0.049) 944 0.192

0.055 (0.050) 934 0.225

0.225⁎⁎⁎ (0.070) 0.511 (1.749) −0.010 (0.065)

769 0.245

1.308 (2.141) −0.039 (0.080) −3.308⁎ (1.963) 0.129⁎ (0.075) 1.668 (3.225) −0.046 (0.120)

−3.518 (2.486) 0.136 (0.095) 2.915 (3.848) −0.090 (0.144)

777 0.194

0.184⁎⁎ (0.074)

−0.008 (0.033) 944 0.191

−0.001 (0.036) 934 0.224

0.205⁎⁎⁎ (0.066) 0.518 (1.746) −0.010 (0.065)

−3.278⁎ (1.966) 0.128⁎ (0.075) 1.647 (3.226) −0.045 (0.120)

−3.606 (2.632) 0.140 (0.101) 3.176 (3.860) −0.099 (0.144)

715 0.195

0.168⁎⁎ (0.070)

707 0.252

0.018 (0.029) 944 0.191

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

Leverage

Intensive margin (existing exporters who increased export/sales)

0.021 (0.031) 934 0.225

Note: The dependent variable is firm sales growth between 1997 and 1998 and the independent variables are for year 1996. Firms with negative net worth are excluded from the sample. Robust standard errors are reported in parentheses. All regressions include industry dummies at the two-digit level and the lagged sales growth rate. ⁎⁎⁎ Denotes a p-value less than 1%. ⁎⁎ Denotes a p-value less than 5%. ⁎ Denotes a p-value less than 10%.

227

228

Table A4 Fixed-effect panel regressions. Publicly-listed firms

Full sample Sales growth

Size Leverage Leverage ∗ crisis ST debt ratio ST debt ratio ∗ crisis

Export/sales ∗ crisis FC debt ratio FC debt ratio ∗ crisis

1

2

3

4

5

6

7

8

9

10

11

12

−0.383⁎⁎⁎ (0.045) 0.765⁎⁎⁎

−0.386⁎⁎⁎ (0.045) 0.752⁎⁎⁎

−0.388⁎⁎⁎ (0.047) 0.746⁎⁎⁎

−0.472⁎⁎⁎ (0.092) 0.450 (0.282) −0.149 (0.149) 0.234 (0.248) 0.051 (0.154)

−0.476⁎⁎⁎ (0.093) 0.437 (0.284) −0.152 (0.150) 0.231 (0.248) 0.060 (0.157)

−0.483⁎⁎⁎ (0.097) 0.399 (0.286) −0.159 (0.154) 0.264 (0.254) 0.101 (0.174)

−0.689⁎⁎⁎ (0.090) −0.474⁎

(0.140) −0.216⁎⁎⁎ (0.075) −0.194⁎ (0.106) 0.001 (0.098)

−0.744⁎⁎⁎ (0.056) −0.202 (0.152) −0.120⁎ (0.066) 0.052 (0.124) 0.159 (0.101)

−0.721⁎⁎⁎ (0.096) −0.499⁎⁎

(0.138) −0.221⁎⁎⁎ (0.074) −0.198⁎ (0.104) 0.035 (0.093)

−0.757⁎⁎⁎ (0.056) −0.220 (0.150) −0.105 (0.066) 0.083 (0.122) 0.146 (0.096)

−0.712⁎⁎⁎ (0.095) −0.460⁎

(0.137) −0.222⁎⁎⁎ (0.074) −0.196⁎ (0.104) 0.030 (0.093)

−0.754⁎⁎⁎ (0.056) −0.202 (0.150) −0.105 (0.066) 0.085 (0.122) 0.140 (0.096)

(0.244) −0.168 (0.137) 0.266 (0.226) 0.452⁎⁎⁎

(0.245) −0.191 (0.138) 0.248 (0.226) 0.499⁎⁎⁎

(0.244) −0.213 (0.143) 0.165 (0.222) 0.473⁎⁎⁎

(0.150)

(0.150)

(0.165)

−0.142⁎⁎⁎

−0.138⁎⁎⁎

−0.137⁎⁎

−0.132⁎⁎

(0.053) 0.244⁎⁎⁎ (0.062)

(0.053) 0.234⁎⁎⁎ (0.062)

−0.119⁎⁎ (0.054) 0.235⁎⁎⁎ (0.063)

−0.104⁎ (0.061) 0.161⁎⁎⁎ (0.058)

−0.063 (0.061) 0.219⁎⁎⁎ (0.075)

−0.060 (0.061) 0.212⁎⁎⁎ (0.076)

−0.009 (0.026) 0.024 (0.025)

−0.096 (0.417) −0.539⁎⁎

Size ∗ FC debt Size ∗ FC debt ∗ crisis

−0.076 (0.061) 0.204⁎⁎⁎ (0.076)

(0.270) 0.004 (0.017) 0.023⁎⁎

ST FC debt ratio ∗ crisis Size ∗ ST FC debt Size ∗ ST FC debt ∗ crisis LT FC debt ratio LT FC debt ratio ∗ crisis Size ∗ LT FC debt Size ∗ LT FC debt ∗ crisis 15465 0.138 5395

15465 0.139 5395

−0.100 (0.070) 0.349⁎⁎⁎ (0.062)

−0.098 (0.071) 0.347⁎⁎⁎ (0.061) −0.479 (0.799) −0.027 (0.520) 0.020 (0.031) 0.004 (0.020)

−0.157⁎⁎⁎ (0.028) 0.146⁎⁎⁎

−0.119 (0.571) −0.704⁎⁎

−0.003 (0.054) 0.082⁎

(0.023)

(0.321) −0.001 (0.024) 0.035⁎⁎⁎

(0.049)

(0.011) ST FC debt ratio

Observations R-squared Number of firms

I/K−1

(0.013)

(0.049)

−1.018 (1.370) −0.932 (0.636) 0.033 (0.054) 0.040 (0.025)

(0.055) 0.011 (0.832) −0.782 (0.608) −0.012 (0.035) 0.0417⁎ (0.025)

−0.628 (1.268) 0.266 (0.841) 0.031 (0.050) −0.009 (0.033)

0.316 (2.083) −1.078 (1.351) −0.031 (0.083) 0.057 (0.053)

−0.504 (0.634) 0.094 (0.533) 0.024 (0.027) −0.004 (0.022) 15214 0.122 5288

−0.238⁎⁎⁎ (0.072) 0.111⁎⁎

(0.061) 0.179⁎⁎⁎ (0.059)

0.433 (1.569) −1.226 (1.440) −0.012 (0.060) 0.047 (0.056)

1.657 (1.739) −2.695⁎⁎ (1.373) −0.065 (0.070) 0.112⁎⁎

15214 0.121 5288

(0.061) 0.193⁎⁎⁎ (0.061)

−0.078 (0.826) 0.168 (0.917) 0.000 (0.032) −0.004 (0.035)

0.236 (1.305) −1.470 (1.301) −0.013 (0.052) 0.065 (0.051)

15330 0.140 5375

−0.095 (0.072) 0.348⁎⁎⁎ (0.062)

15082 0.121 5271

3750 0.216 1104

3750 0.216 1104

3711 0.218 1101

3729 0.162 1098

3729 0.166 1098

3690 0.159 1096

Note: The dependent variables are the sales growth rate and the ratio of investment to lagged capital. The independent variables are lagged by one year. Firms with negative net worth are excluded from the sample. Crisis refers to a dummy which takes one in 1998 and zero otherwise. Individual- and time-fixed effects are included. FC debt denotes net foreign currency denominated debt. Robust standard errors are reported in parentheses. ⁎⁎⁎ Denotes a p-value less than 1%. ⁎⁎ Denotes a p-value less than 5%. ⁎ Denotes a p-value less than 10%.

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

Export/sales

Sales growth

I/K−1

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

229

Table A5 Balanced-sample regressions for crisis. Publicly-listed firms

Full sample 1 Chaebol dummy Age Size

0.008 (0.022) 0.001 (0.001) −0.018⁎⁎⁎ (0.005)

Leverage

0.004 (0.050)

Size ∗ leverage ST debt ratio

0.033 (0.061)

Size ∗ ST debt ratio Export/sales

0.184⁎⁎⁎ (0.033)

Foreign debt ratio

0.050⁎⁎ (0.020)

Size ∗ foreign debt

2 −0.002 (0.024) 0.001 (0.001) 0.005 (0.023)

−0.001 (0.024) 0.001 (0.001) 0.005 (0.023)

0.628 (0.913) −0.026 (0.038) 1.169 (1.168) −0.048 (0.049)

0.713 (0.906) −0.030 (0.038) 0.935 (1.154) −0.037 (0.049)

0.180⁎⁎⁎ (0.032)

0.180⁎⁎⁎ (0.032)

−0.786⁎⁎ (0.341) 0.035⁎⁎ (0.014)

4

5

6

0.026 (0.029) 0.002⁎

0.030 (0.030) 0.002⁎

0.022 (0.029) 0.002⁎

(0.001) −0.020⁎⁎ (0.010)

(0.001) 0.044 (0.047)

(0.001) 0.041 (0.047)

2.367 (2.010) −0.096 (0.079) 0.123 (1.692) −0.003 (0.067)

2.252 (2.048) −0.093 (0.081) 0.256 (2.010) −0.005 (0.080)

0.004 (0.094)

0.038 (0.091)

0.154⁎⁎⁎ (0.055) 0.101⁎ (0.057)

0.152⁎⁎⁎ (0.056)

Size ∗ ST foreign debt LT foreign debt ratio Size ∗ LT foreign debt 2633 0.128

2633 0.131

2633 0.131

0.146⁎⁎⁎ (0.056)

−0.336 (0.846) 0.018 (0.032)

−0.266 (0.553) 0.012 (0.022) −1.247⁎ (0.670) 0.055⁎⁎ (0.028)

ST foreign debt ratio

Observations R-squared

3

−1.971 (1.231) 0.075 (0.047) 0.049 (1.683) 0.007 (0.066) 777 0.219

777 0.223

777 0.236

Note: The dependent variable is firm sales growth between 1997 and 1998. The independent variables are for year 1996. Firms with negative net worth are excluded from the sample. Robust standard errors in parentheses. All regressions include industry dummies at the two-digit level and the lagged sales growth rate. ⁎⁎⁎ Denotes a p-value less than 1%. ⁎⁎ Denotes a p-value less than 5%. ⁎ Denotes a p-value less than 10%.

Fig. A1. Coefficients of foreign debt and IICs across industries. Note: IIC (imported input coefficient) is the share of imported inputs in total intermediate inputs for each two-digit industry. β1 + β2 ∗ size is the estimated coefficient of foreign debt on sales growth for each two-digit industry, evaluated at the 10th percentile of firm size.

constructed as above and the economic depreciation rate d is assumed to be 11%, which is an average depreciation of building, structure, vehicle and machine in South Korea. The initial capital stock is measured as the real book value of capital in the year that a firm first appears in the data set. References Aguiar, M., 2005. Investment, devaluation, and foreign currency exposure: the case of Mexico. J. Dev. Econ. 78, 95–113.

Allayannis, G.I., Ofek, E., 2001. Exchange rate exposure, hedging, and the use of foreign currency derivatives. J. Int. Money Financ. 20, 273–296. Altman, E.I., 1968. Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. J. Financ. 23 (4), 589–609.

230

Y.J. Kim et al. / Journal of International Economics 97 (2015) 209–230

Altman, E.I., 2000. Predicting financial distress of companies: revisiting the z-score and ZETA models. Stern School of Business Working Paper. New York University. Bae, S.C., Kwon, T.H., 2010. Asymmetric foreign exchange exposure and foreign currency denominated debt: international evidence. Working Paper. Bowling Green State University and Chungnam National University. Bayraktar, N., Sakellaris, P., Vermeulen, P., 2005. Real versus financial frictions to capital investment. European Central Bank Working Paper 566. Benavente, J.M., Johnson, C.A., Morande, F.G., 2003. Debt composition and balance-sheet effects of exchange rate depreciations: a firm-level analysis for Chile. Emerg. Mark. Rev. 4 (2003), 368–396. Bernanke, B.S., Gertler, M., Gilchrist, S., 1999. The financial accelerator in a quantitative business cycle framework. In: Taylor, J.B., Woodford, M. (Eds.), Handbook of Macroeconomics. Elsevier, pp. 1341–1393. Bleakley, H., Cowan, K., 2008. Corporate dollar debt and depreciations: much ado about nothing? Rev. Econ. Stat. 90 (4), 612–626. Bleakley, H., Cowan, K., 2009. Mishmash on mismatch? Balance-sheet effects and emerging-markets crises. Working Paper. University of Chicago. Bonomo, M., Martins, B., Pinto, R., 2003. Debt composition and exchange rate balancesheet effect in Brazil: a firm level analysis. Emerg. Mark. Rev. 4, 368–396. Borensztein, E., Lee, J.W., 2002. Financial crisis and credit crunch in Korea: evidence from firm-level data. J. Monet. Econ. 49, 853–875. Carranza, L.J., Cayo, J.M., Galdon-Sanchez, J.E., 2003. Exchange rate volatility and economic performance in Peru: firm level analysis. Emerg. Mark. Rev. 4, 472–496. Céspedes, L.J., Chang, R., Velasco, A., 2004. Balance sheets and exchange rate policy. Am. Econ. Rev. 94 (4), 1183–1193. Corsetti, G., Pesenti, P., Roubini, N., 1999. What caused the Asian currency and financial crisis? Jpn. World Econ. 11 (3), 305–373. Echeverrya, J.C., Fergussona, L., Steinerb, R., Aguilara, C., 2003. Dollar debt in Colombian firms: are sinners punished during devaluations? Emerg. Mark. Rev. 4, 417–449. Fazzari, S., Hubbard, R.G., Peterson, B., 1988. Financing constraints and corporate investment. Brook. Pap. Econ. Act. 1, 141–195. Feldstein, M., 1999. Self-protection for emerging market economies. NBER Working Paper 6907. Forbes, K.J., 2002. How do large depreciations affect firm performance? IMF staff papers 49.

Geczy, C., Minton, B.A., Schrand, C., 1997. Why firms use currency derivatives. J. Financ. 52, 132–1354. Gertler, M., Gilchrist, S., 1994. Monetary policy, business cycles, and the behavior of small manufacturing firms. Q. J. Econ. 109 (2), 309–340. Gertler, M., Hubbard, R.G., 1988. Financial factors in business fluctuations. Financial Market Volatility: Causes, Consequences, and Policy. Federal Reserve Bank of Kansas City, Kansas City, pp. 33–71. Gilchrist, S., Sim, J.W., 2007. Investment during the Korean financial crisis: a structural econometric analysis. NBER Working Paper 13315. Grubaugh, S.G., 1987. Determinants of direct foreign investment. Rev. Econ. Stat. 69 (1), 149–152. Horst, T., 1972. Firm and industry determinants of the decision to invest abroad: an empirical study. Rev. Econ. Stat. 54 (3), 258–266. Jeon, Y.D., 1992. The determinants of Korean foreign direct investment in manufacturing industries. Rev. World Econ. 128 (3), 527–542. Kalemli-Ozcan, S., Kamil, H., Villegas-Sanchez, C., 2010. What hinders investment in the aftermath of financial crises: insolvent firms or illiquid banks? NBER Working Paper 16528 Krugman, P., 1999. Balance sheets, the transfer problem, and financial crises. Int. Tax Public Financ. 6 (4), 459–472. Lee, J.W., Lee, Y.S., Lee, B.S., 2000. The determination of corporate debt in Korea. Asian Econ. J. 14, 333–356. Luengnaruemitchai, P., 2003. The Asian crisis and the mystery of the missing balancesheet effect. Working Paper. University of California, Berkeley. Martínez, L., Werner, A., 2002. The exchange rate regime and the currency composition of corporate debt: the Mexican experience. J. Dev. Econ. 69 (2002), 315–334. Min, B., 2007. Changing pattern of corporate governance and financing in the Korean chaebols. Econ. Pap. 26 (3), 211–230. Pratap, S., Lobatoa, I., Somuanob, A., 2003. Debt composition and balance-sheet effects of exchange rate volatility in Mexico: a firm level analysis. Emerg. Mark. Rev. 4, 450–471. Zingales, L., 1998. Survival of the fittest or the fattest? Exit and financing in the trucking industry. J. Financ. 53 (3), 905–938.