Financial Shocks and Trade Finance: Evidence from Korea * Sangyeon Hwang Kongju National University & Hyejoon Im Yeungnam University March 2012 * We would like to thank Joonmo Kwon at the Bank of Korea for his extremely helpful guides for trade finance instruments in Korea and their data. Also, we thank Youngjoon Kim and the seminar participants at the Sixth Joint Seminar of Hokkaido and Yeungnam University and the 2012 Economics Joint Conferences in Korea for their useful comments. All errors herein are ours. Assistant Professor, Department of Economics and Trade, Kongju National University, 182 Singwan-dong, Gongju 314-701, South Korea; Phone: +82-41-850-8391; Fax: +82-41-850-8389; Email: syhwang@kongju.ac.kr. Corresponding author; Assistant Professor, School of Economics and Finance, Yeungnam University, Kyungsan 712-749, South Korea; Phone & Fax: +82-53-810-2846; Email: hi6w@ynu.ac.kr.
Financial Shocks and Trade Finance: Evidence from Korea March 2012 Abstract We empirically assess the impacts of financial shocks on trade finance. Unlike previous studies, we use two direct measures for trade finance in Korea: foreign trade loans extended by commercial banks and documentary bills bought by them such as LCs and D/P or D/A. Both are representative bank-intermediate instruments of trade finance in Korea. By estimating a vector autoregression (VAR), we find that regardless of which measures are used, the impacts of financial shocks are negative but short-lived. Although sensitive to lag-specification, we also found, that the past two financial crises adversely affected trade finance (measured by foreign trade loans) via a different channel. High exchange rate volatility or country risk was a driving force of the deterioration of trade finance during the Asian financial crisis in 1997, whereas the global credit constraint was a main contributing factor to its deterioration during the recent global financial crisis. JEL codes: G01, G21, F10, F40 Keywords: trade finance, financial crisis, loan, documentary bill 1
1. Introduction As the collapse of world trade began to be felt as a consequence of the global financial crisis, Pascal Lamy, the Director-General of the World Trade Organization, warned on November 12 th 2008 that the market for trade finance has severely deteriorated over the last six months. He pointed out two key causes for this. One was a shortage of liquidity to trade financing. The other was a general re-assessment of risks caused as much by the financial crisis as by the slowing down of the world economy. He further mentioned that traders and banks in the emerging market economies most severely experience these problems, and that if trade finance is not tackled we run the risk of further exacerbating the downward spiral of world trade and economy (WTO, 2008). In this study, we examine the impacts of financial shocks on trade finance, which is traditionally perceived as relatively secure financial products due to the short maturity and collateral status of the traded goods. In so doing, we use rare data of trade finance and employ an unrestricted vector autoregression (VAR) with four financial variables: trade finance, domestic and global credit availability, and country risk. We examine a generalized impulse response of trade finance to financial shocks and then investigate the effects of the financial variables on trade finance particularly in periods of financial crises. In the literature, it is well-known that there are no reliable data on trade finance. For instance, Amiti and Weinstein (2011) point out that the standard proxies for trade finance often used in the literature such as trade credit or short-term credit are indeed not the right measures. 1 1 Also, refer to IMF (2003) and Chauffour and Farole (2009) for the discussions on the data limitation of trade finance. 2
Trade credit, a term in accounting, is a direct extension of credits between parties: when a firm receives an order for goods or services that will be paid later or are paid in advance, it records a trade credit on the accounts receivable or payable section of its balance sheet. In contrast to previous studies, we use two direct measures for trade finance in Korea: foreign trade loans extended by commercial banks and documentary bills they bought, such as Letter of Credits (LCs) and Documents against Payment (D/P) or Documents against Acceptance (D/A). 2 Both are typical bank-intermediate instruments of trade finance not only in Korea but also in other countries. Banks can extend foreign trade loans to exporters so that companies have sufficient working capital for the pre-shipment period of the goods. Since these loans are only extended based upon past or current exporting activities of firms, they should be distinguished from other general bank loans. On the other hand, after shipping the goods, exporters can ask banks to accept documentary bills so that they have funds in advance of importers payment. We find that the impacts of financial shocks are adverse but short-lived. When we use foreign trade loans as a measure for trade finance, we find that shortages in both domestic and global credit availability, as well as an increase in country risk, can deteriorate trade finance in Korea but these impacts are short-lived (about two months). In our study, domestic and global credit availability are measured by a spread between a treasury bond rate and a corporate bond rate in Korea and a spread between a U.S. T-bill rate and a LIBOR rate, respectively, and country risk is measured by exchange rate volatility. Similarly, when we use documentary bills as a measure for trade finance, their impulse responses to other variables are negative, whenever statistically significant, and the persistence of the impulse responses is brief (just one quarter). 2 We explain various trade instruments in detail in Section 3. 3
In the case of foreign trade loans, we find that the impacts of financial shocks on trade finance are magnified in periods of crises, although the result is sensitive to lag-specification. Furthermore, depending on the characteristics of financial crisis episodes, it is noteworthy that variables that exhibit particularly stronger negative relationships with trade finance during a crisis period can differ. A large swing of the exchange rate or a high country risk was a driving force of the deterioration of trade finance during the Asian financial crisis in 1997, while the global credit constraint was a major source of deterioration during the recent global financial crisis. The paper is organized as follows. Section 2 reviews the previous literature on trade finance. Section 3 overviews various instruments of trade finance in Korea. Section 4 explains data in use and provides an empirical framework. Section 5 presents the results on the generalized impulse responses of trade finance to financial shocks and investigates the impacts on trade finance during financial crises. Section 6 summarizes the findings and contains some concluding remarks. 2. Literature review Even though it has played a paramount role in facilitating trade, trade finance has been paid very little attention in the literature. While there are some literatures that study the relationship between trade finance and international trade during financial crises, few existing studies examine the effects of financial shocks or crises on trade finance,. A survey commissioned by the International Monetary Fund and the Bankers 4
Association for Trade and Finance reports that the values of LCs, export credit insurance and short-term export working capital business have decreased by 3-11% over the period of October 2008 through January 2009 (FImetrix, 2009). 3 The respondents in the survey attribute the decrease in values of trade transactions to a fall in the demand for trade activities followed by less credit availability at both domestic and foreign financial institutions. The survey also reports the increase, sometimes doubling, in the price of trade finance instruments such as LCs and export credit insurance, which was mainly due to the increased costs of funds for financial institutions. The International Chamber of Commerce (2009) reports, based on a survey of banks in many countries, that open account trading experienced a decrease while demand for LCs increased over the last quarter of 2008, compared with the same period in 2007. Related to this, it is important to note that most respondents rate the risk associated with traditional trade instruments like LCs as same as, or significantly lower than, the risk for general banking instruments such as open accounts. Using a sample of firms in six emerging economies, Love et al. (2007) examine the effects of past financial crises on trade credits between parties in cross-border transactions. They find that the provision of trade credits increases right after a crisis but reduces in the period following the crisis. Although it is interesting to investigate how firms adjust the extension of credits between parties in response to financial shocks, the current study is more interested in the impacts on trade finance instruments that are intermediated by banks or financial institutions. Chauffour and Farole (2009) qualitatively assess trade finance in the recent financial crisis, and assert that a decrease in trade finance leads to a drop in international trade, but assert 3 See the next section for explanation on the various instruments of trade finance. 5
that its contribution to the massive decline in world trade should not be overestimated. They also discuss whether the decline in the supply of trade finance is a consequence of market or government failure, and thus whether there is justification for public intervention in the trade finance market. On the other hand, several studies have mainly looked at a different issue, though related, which is the effects of trade finance on trade during financial crises. Ronci (2004) assesses the effect of trade finance on trade flows in countries in periods of financial crises. Using outstanding short-term credit as a proxy for trade finance, he finds that a fall in trade finance explains only a small part of the trade loss during crises. As described earlier, however, what he uses as a proxy for trade finance may not be the correct measure. Short-term credit may be an instrument of trade finance in a broad sense, but it should be distinguished from trade finance instruments that involve bank or other financial institutions such as LCs, loans, and guarantees. Given the paucity of data on trade finance, Amiti and Weinstein (2011) use a dataset that allows matching exporters with the main banks that they had banking businesses with, including trade finance, during the Japanese financial crises of the 1990s. They find that availability of trade finance explains as much as one-third of the fall in Japanese exports during the crises. However, they do not use a direct proxy for trade finance, since the availability of trade finance in their analysis can be inferred only by the healthiness of the banks matched with the exporters. Levchenko et al. (2010) examine the likely causes for the collapse of international trade during the recent global crisis, including trade credits as a proxy for trade finance, and find that trade credits do not play a role in the recent trade collapse. In contrast, Ahn et al. (2011) provide some evidence that trade finance may have played a role in a greater decline in exports during the recent crisis. They show that export prices rose relative to domestic manufacturing prices 6
during the crisis, and the prices of seaborne imports and exports which are likely to be affected most by trade finance contraction rose relative to goods sent by land or air. In the current study we concentrate on the effects of financial shocks on trade finance and, unlike previous studies, directly measure trade finance instruments that rarely have data available. We now look at this measure in further detail while considering a variety of trade finance instruments in Korea. 3. The overview of trade finance in Korea Various trade finance instruments in Korea are summarized in Table 1. Depending on the types of intermediary, we can categorize trade finance instruments into three types: bankintermediate instruments, other financial institute-intermediate instruments, and a direct extension of credit between traders. 4 These instruments are typical in the trade finance market in other countries as well. 5 [Table 1 insert here] In a narrow sense, trade finance in Korea is usually referred to as an extension of loans to exporters before shipping (Song, 1999). This loan is called a foreign trade loan. 6 For the 4 It is important to distinguish the methods of payments for cross-border transactions from the trade finance instruments. Typically, exporters are paid by one of the three methods: LCs, D/P or D/A, or open accounts. 5 Refer to Auboin and Meier-Ewert (2003) for a general description of various instruments of trade finance. 6 The Bank of Korea maintains a loan program, which is designed to help small- and medium-sized companies and 7
period of pre-shipment of the goods, commercial banks may extend loans to exporters so that companies have sufficient working capital. Since these loans can only be extended based either upon LCs issued to exporters or the past export performance of exporters, they can be distinguished from other general bank loans. 7 Another bank-intermediate form of trade finance is one of documentary bills, bought by banks after exporters ship goods. The importers banks (issuing banks) issue LCs to exporters to assure that exporters are paid when they submit the required documents (invoices, bills of lading, etc.) as stipulated on LCs to their banks (the advising or confirming banks). After shipping the goods, exporters can ask their banks to accept documentary bills attached to LCs, which is called negotiation, so that they receive payment in advance of when importers pay the bills. Simply put, LCs are a conditional bank undertaking of payment on behalf of importers such that the issuing banks make payment upon the presentation of stipulated documents (at sight LCs) or at a later specified date (usance LCs). Documentary collection such as Documents against Payment (D/P) and Documents against Acceptance (D/A) is another instrument of bank-intermediate trade finance in the postshipment period and involves documentary bills. When exporters ship goods on a documentary collection basis, they draw a draft (bill of exchange) on the importer for the total amount due, attach the draft to shipping documents, and submit it to their banks for collection from the importer. Unlike LCs, banks in documentary collection simply act as an agent for the exporters to promote balanced regional development. It sets a maximum amount of loans for this program in a periodic manner, which is distributed to commercial banks at a lower interest rate than the market rate. Some portions of these loans are allocated for the extension of foreign trade loans. 7 Another trade finance instrument in the pre-shipment period is trade bills discounted by banks. However, this instrument is far less popular for both exporters and banks due to the procedural complexity relative to foreign trade loans. 8
and do not have any responsibility to make payments. Exporters often ask their banks to simply accept bills and to make payments in advance of collection. Next, export credit insurance or guarantees are the instruments that insure or guarantee exporters against non-payment, political risk, and conflicts. The Korean Trade Insurance Corporation and the Korea Export-Import Bank, both of which are sponsored by the government, are responsible for such instruments. Finally, trade credit is a direct extension of credits between parties, which does not involve banks or financial institutions. Companies extend credit to the other party when buyers delay or advance payments to suppliers. Exporters may extend credit to importers such that they are paid later when goods are received by importers. In contrast, importers may extend credit to exporters such that importers pay for the goods in advance, before shipping. 8 Among various instruments of trade finance, we are primarily interested in the impacts of financial shocks on the supply of trade finance that are intermediated by commercial banks. Bank-intermediate trade finance is very different from a direct extension of trade credits between parties. It is also an operation of private financial business and thus more responsive to financial crisis events compared with export credit insurance or guarantee offered by public institutions. As described earlier, there are two bank-intermediate instruments depending on the period of shipment of goods: one is foreign trade loans extended by commercial banks in the pre-shipment and the other is documentary bills such as LCs and D/P (D/A) purchased in the post-shipment period. We use data of both instruments as a proxy for trade finance in Korea. 9 To our 8 Trade credits are usually implemented by using the following methods of payments: open accounts, Cash on Delivery (COD), and Cash against Documents (CAD) or European D/P. 9 Data for LCs and D/P (D/A) are only available together, not separately, since both instruments basically involve documentary bills. 9
knowledge, this study is the first to use these measures, which we believe are the closest possible proxies for trade finance, compared with other measures used in past studies Let us briefly describe the trend of the balance of foreign trade loans and documentary bills that are extended or bought by commercial banks. The Bank of Korea keeps track of the monthly balance of foreign trade loans by collecting data from commercial banks. (See Figure 1.) The annual average growth rate of the balance for this period is recorded as 9.5%. The balance was USD 3,661 million on January 1, 1996, and declined sharply during the Asian crisis. One observation is that on May 1998, the Bank of Korea removed an upper limit for which commercial banks could supply foreign trade loans, which then led to a surge in the following couple of years. 10 Since then, the balance steadily increased before the decline during the recent global crisis. It almost recovered to the pre-crisis level on March 2010, and recorded USD 11,311 million as of December 31, 2010. [Figure 1 insert here] The Financial Supervisory Service of Korea collects quarterly data on the balance of documentary bills. (See Figure 2.) As of 1999Q1, the balance was USD 15,004 million. The balance increased to the record high of USD 30,100 million in 2008Q3, about twice as large as the balance in 1999Q1. From the peak in 2008Q3, when Lehman Brothers went into bankruptcy, the balance began to collapse and decreased by 40.9% to the level of USD 17,777 million in 2009Q3. Since then, the balance rebounded to the level of USD 22,449 million in 2010Q4. 10 According to a news report (MK Business News, April 11th 1998), the upper limit for foreign trade loans was abolished on May of that year, in order to reduce the difficulties for exporters due to the shortage of trade finance, and thus commercial banks could freely choose their own limits for the loans. 10
[Figure 2 insert here] 4. Data and empirical methodology We estimate an unrestricted vector autoregression (VAR) model with four variables: trade finance, domestic and global credit availability, and country risk. We have chosen these particular financial variables based on the following observations. Auboin and Meier-Ewert (2003) provided two important implications of the financial crises in the 1990s for emerging market economies. First, the exchange rate became so volatile during the crises that this exacerbated the economic fundamentals and further worsened the problem of capital flight. Second, there was a scarcity of short-term trade-financing facilities in the period of crisis, which made things difficult for both exporters and importers. Similarly, as described in the introduction, Lamy succinctly stated that liquidity shortages and a general re-assessment of risks of the economy were the main causes for the deterioration of trade finance in the period of crisis (WTO, 2008). 11 Given the above discussion, we have chosen exchange rate volatility and both domestic and global credit availability as key variables that may be related to trade financing conditions. As described earlier, we use both foreign trade loans and documentary bills as a measure of trade finance in Korea. It is likely that larger exports create more demand for trade finance, and vice versa for smaller exports. As we are primarily interested in the supply side of trade finance in response to shocks faced by the finance market, we need to control the demand side of 11 This is an official stance of the WTO about the issue. See Auboin (2009). 11
trade finance. Hence, we use a ratio of foreign trade loans or documentary bills to a 12-month rolling sum of exports, which we denote as TFBOK_EX and TFBILL_EX, respectively. 12 In a sense, these ratios represent the supply side of financing export activity, while controlling for the demands of trade finance in association with export activity. 13 The trends of the two ratios are shown in Figure 3 and Figure 4. 14 [Figures 3 & 4 insert here] TED spread is used as a proxy variable for global liquidity constraint in response to a shock to the global banking system or a perceived credit risk in the global financial market (Bijapur, 2010). 15 TED spread is the difference between a U.S. three-month Treasury bill rate and a three-month LIBOR rate for Eurodollar. In other words, it measures the credit risk premium between a secure government bond rate and an unsecured deposit rate. The increase in TED spread indicates shortages of global credit availability. Hence, we predict a negative relationship between TED spread and trade finance. 12 The reason why we use a 12-month rolling sum of exports is as follows. The balance of foreign trade loans is a stock variable calculated at the end of the month, while that of exports is a flow variable measured for the corresponding month. Hence, we need to make the two variables comparable. In addition, given the fact that the extension of these loans is based on LCs or past export performances, the flows of past exports are likely to be correlated with the current balance of foreign trade loans. 13 In the examination of the effects of financial crises on the extension of trade credits between parties, Love et al. (2007) similarly scale trade credits (accounts receivables) using sales to control for declines in economic activity that are common during crises. 14 One observation is that the ratio of documentary bills to exports declines over the sample period. This trend reflects the trend of the method of payments of exports. Korea s exports transacted via open accounts have risen from 40.1% in January 1999 to 59.5% in January 2009, while those transacted via documentary bills have fallen from 54.5% to 24.6% over the same period. 15 Cheung et al. (2010) also state TED spread adjusts to new information rapidly and serves as a leading fear indicator, not only for the US market but also for other global markets. 12
By analogy with the global credit condition, we introduce a domestic counterpart. We use a spread between a three-year Korean treasury bond rate and a three-year corporate bond rate (AA- rate) in Korea, which we call Korean spread from here on (KSPRD). Higher Korean spread implies less availability of domestic credit. Again, we expect a negative association between Korean spread and trade finance. As seen in the Asian crisis in 1997, the large swings in the exchange rate exacerbate the fundamental weakness of the economy including financial fragility, external vulnerability, and poor governance (Auboin and Meier-Ewert, 2003). Thus, we use exchange rate volatility as a proxy for weak economic fundamentals or country risk. Exchange rate volatility is defined as a standard deviation of the bilateral nominal exchange rate between Korean Won and U.S. dollar, divided by its period average (FXSTD). Higher exchange rate volatility leads to higher country risk, which suggests a negative relationship between exchange rate volatility and trade finance. The VAR model with foreign trade loans is implemented on a monthly basis. The data of the balance of foreign trade loans, TED spread, exchange rate, and Korean spread are all obtained from the Bank of Korea. We obtain data on exports from the Korea International Trade Association. The sample is monthly from January 1996 to December 2010. 16 On the other hand, the data of the balance of documentary bills bought by commercial banks are only available on a quarterly basis from 1999Q1 to 2010Q4. We obtain them from the Financial Supervisory Service of Korea. When we implement the VAR analysis with documentary bills, we use the quarterly data for other variables by taking a simple average of their monthly data. One caveat with the 16 Although the data of foreign trade loans, TED spread, and exchange rate are available for an earlier period, the data of Korean spread is only available from May 1995. 13
data of documentary bills is that they have a short-time span and a low- time frequency, which might not be ideal for an analysis of financial variables. To examine the relationship between variables in the VAR analysis, we use the generalized impulse response function. Unlike the conventional impulse response function, the generalized impulse response provides robust results to the ordering of variables (Pesaran and Shin, 1998). Consider the VAR model, which is similar to the one employed by Pesaran and Shin (1998), as shown here: x c Φ x ε, t 1,2,,T (1) where i and t denote the number of lags and time, respectively, x x,x,,x is an m 1 vector of endogenous variables, Φ is an m m coefficients matrix, and ε is a vector of white noise. The vector of moving average representation of equation (1) under a standard stability assumption 17 is: x μ Ψ ε, t 1,2,,T (2) where Ψ, an m m coefficients matrix, can be obtained using the following recursive relations: Ψ =Φ Ψ Φ Ψ Φ Ψ, i 1,2,3, (3) 17 All values of z satisfying I Φ z 0 fall outside the unit circle. 14
where Ψ I and Ψ 0 for i 0. In this model, the orthogonalized impulse response function of x at time t+n, x, with respect to a unit shock in the jth equation is given by: ξ n Ψ Pe, n 0,1,2, (4) where P is an m m lower triangular matrix from the Cholesky decomposition of, the covariance matrix of ε, and e is an m 1 selection vector with unity at the jth element and zeros elsewhere. The generalized impulse response function for x with respect to one standard error shock to the jth equation at time t is given by: ξ n σ / Ψ Σe, n 0,1,2, (5) where σ is the error variance of the jth equation (j-j element of ). It is easy to show that the generalized impulse response equation (5) is invariant of the ordering of the variables in the VAR. This is in contrast to the orthogonalized impulse response equation (4), which is affected by alternative choices of decompositions of, and thus the ordering of the variables included in the VAR model. In the following analyses, we use this generalized impulse response equation to examine relationships between variables. The goal of the analysis is to measure the effects of financial shocks on trade finance. For that, we proceed in the following manner. First, we analyze the generalized impulse response 15
to assess how a shock to financial variables affects trade finance. Second, we use a conventional VAR setting and introduce a dummy variable for crisis periods that interacts with each variable other than trade finance. Then, the coefficients of the interactive dummy variables can be regarded as estimates of the impact of each variable on trade finance in periods of financial crises. 5. Results and discussions 5. 1. Generalized impulse response of trade finance We estimate a VAR model such as equation (1) with the four variables: TED spread (TED), exchange rate volatility (FXSTD), Korean spread (KSPRD), and the trade finance condition (TFBOK_EX or TFBILL_EX). Before estimating the VAR, we perform unit-root tests and find that Augmented Dickey-Fuller, Dickey-Fuller GLS, and Ng-Perron tests indicate that all variables, except both TFBOK_EX and TFBILL_EX, are stationary. Hence, these variables are first-differenced to attain stationarity. As described earlier, the ceiling on foreign trade loans for commercial banks was removed on May 1998. Hence, we control this policy shift by introducing a time dummy for the post-1998m4 observations in the VAR with foreign trade loans. 18 We describe the results for foreign trade loans and documentary bills, in turn, in the following. 18 We have also estimated the model for the period of 1999m1-2010m12, which excludes the policy shift. We find a statistically significant negative relationship between trade finance and TED spread or exchange rate volatility, but not for Korean spread. The persistence of impulse responses is six months for TED spread and three months for exchange rate volatility. 16
When using foreign trade loans (TFBOK_EX) as a measure for trade finance, various lag-length selection criteria suggest either one or two lags. Akaike Information Criterion (AIC), Final Prediction Error (FPE) criterion, and Hannan & Quinn (HQ) criterion suggest two lags, while Schwartz Criterion (SC) suggests one lag. Hence, we estimate the VAR model with each lag. Figure 5 shows the generalized impulse responses of the variables in the specification of one lag. Among them, the impulse response of foreign trade loans with respect to the various shocks is our main interest, as highlighted in the last four panels of Figure 5. We find that the impulse responses of foreign trade loans to other variables are all negative, but only the impulse responses to exchange rate volatility are statistically significant at the 5% level. In the specification of two lags, the negative relationship between foreign trade loans and TED spread, exchange rate volatility, or Korean spread is both clear and significant (the last four panels of Figure 6). Regarding the length of impacts, using the 5% significance level, all impulse responses persist only for two months after the shocks first occur. TED spread has a peak effect on foreign trade loans two months after the shock occurs. In the case of both exchange rate volatility and Korean spread, the peak effect occurs about one month after the shock. Comparing the magnitudes of the peak impulse responses, we find that the impact of exchange rate volatility is the largest, followed by that of Korean and TED spread. 19 19 The impulse responses of other variables in Figure 6 are generally consistent with the economic intuition. For example, the impulse responses of TED spread with respect to any shocks are found to be statistically insignificant, while the impulse responses of exchange rate volatility and Korean spread are significantly positively related with TED spread. This makes sense since TED spread, which measures the global liquidity condition, is unlikely to be affected by the volatility of the Korean currency or the condition of the domestic credit market, while the opposite scenario is likely. 17
[Figures 5 & 6 insert here] When documentary bills (TFBILL_EX) are used as a proxy for trade finance, we have more or less a similar picture. Various lag-length selection criteria suggest either one lag (SC and HQ) or three lags (FPE and AIC). In the one-lag specification, we find that the impulse responses of documentary bills with respect to other variables are all statistically insignificant (the last four panels in Figure 7). On the other hand, in the three-lag specification we find that the impulse responses of documentary bills to TED spread and Korean spread are all negative and significant, while those of exchange rate volatility are negative but insignificant (the last four panels in Figure 8). We also find the negative impulse responses of TED spread and Korean spread persist only for one quarter, which is in line with our earlier finding when using foreign trade loans. [Figures 7 & 8 insert here] In summary, regardless of which measures are used for trade finance, we find that whenever they are statistically significant, the impulse responses of trade finance to financial shocks are negative but short-lived for 2-3 months. In other words, the deterioration in global and domestic liquidity, as well as high exchange rate volatility, adversely affect Korea s trade finance condition, but the impacts are brief. 18
5. 2. The impacts on trade finance during financial crises In this section, we investigate whether the (negative) relationship between trade finance (TFBOK_EX or TFBILL_EX) and the other three financial variables TED spread, exchange rate volatility, and Korean spread becomes strengthened during financial crises. For this, we introduce an interactive dummy variable into the previous VAR model for each variable other than trade finance, with a dummy set equal to 1 for the duration of the crises. Over the sample period for foreign trade loans, there were two major financial crises in Korea the Asian crisis in 1997 and the recent global crisis while there was only the recent crisis over the sample period for documentary bills. We modify the previous unrestricted VAR model as follows: x c Φ x δ CRISIS x, ε, t 1,2,,T (6) where i denotes the number of lags, j indicates a different crisis episode or scheme to be defined shortly, x is the vector of the four variables, δ is the coefficient of an interactive dummy, and x, is one of the three variables in lag (TED, FXSTD, or KSPRD). CRISIS_j is the dummy for a different crisis episode or scheme (j = 1, 2, or 3). CRISIS_1 is set to 1 during the period of the Asian crisis. CRISIS_2 is set to 1 during the period of the recent global crisis. CRISIS_3 is set to 1 during both crisis periods. We follow Yiu et al. (2010) for assigning the time periods for the crises. They define the Asian crisis as the period from October 1997 to December 1998 and the recent global crisis as the period from September 2007 to March 2009. When 19
estimating the model for foreign trade loans, we have included a time dummy for the post- 1998m4 observations to control for a policy shift in the market. In estimation, we introduce an interactive dummy for each variable, one at a time, for each crisis and examine whether the negative relationship between trade finance and a variable in interest is strengthened during a particular crisis. 20 Equivalently, we test the null hypothesis H 0 : δ 0 against the alternative hypothesis H 1 : δ 0 for j = 1, 2, or 3 in the equation of trade finance of the modified VAR system. For illustration, when we test whether the relationship between trade finance and TED spread becomes stronger during the recent global crisis, the interactive terms between the CRISIS_2 dummy and TED spread variable in lag are included in equation (6). If we reject the null, it suggests that the cumulative lagged effects of TED spread on trade finance become larger negatives during the recent global crisis. Given that various lag-selection criteria suggest one or two (three) lags, we estimate the model for foreign trade loans (documentary bills) with either lag. We use the simple t test on the interactive dummy coefficient for one lag, and use the Wald test on the sum of the interactive dummy coefficients for multiple lags. This hypothesis testing also allows us to inspect whether there are any asymmetric relationships between trade finance (foreign trade loans in particular) and other variables across different financial crisis episodes. 21 If such asymmetry is found, this implies that different financial crises affect trade finance through a different channel. In fact, the two financial crises under investigation have different characteristics in terms of their origins and spill-over effects. 20 We could include interactive dummies for all three variables together in equation (6). However, there is a multicollinearity problem in doing this. In fact, when we have tried this specification, most coefficients are found to be statistically insignificant. 21 A similar approach is applied in Bijapur (2010) to investigate the asymmetric effect of monetary policy during a credit crunch. 20
The Asian crisis in 1997 originated in the Asian region and its impacts were largely contained within the region and did not spread globally. Moreover, the crisis was characterized as exhibiting precipitous currency devaluations, leading to a depletion of national holdings of foreign reserves and an increase in country risk. On the other hand, the recent global crisis began in the U.S., and its negative effects spread out globally. In contrast with the Asian one, this crisis was characterized as a liquidity problem in the global financial market. The different characteristics of the two crises can also be seen in Figure 3, which reveals that the three financial variables show different co-movements with trade finance during the two crisis episodes. Specifically, during the period of the Asian crisis, exchange rate volatility and Korean spread increased far more than TED spread. By contrast, an increase in TED spread was most evident in the recent crisis. Therefore, one may expect that there is an asymmetric relationship between trade finance and other variables across the two crisis events, and our findings support this, as described below. Let us first consider foreign trade loans as a measure for trade finance and the specification with one lag. Refer to the first row in Table 2 for the estimation results. The first three columns in Table 2 show the results for the Asian crisis (CRISIS_1). Only the coefficient of the interactive dummy for exchange rate volatility (FXSTD) has a negative sign. According to the hypothesis testing, the null that the coefficient of the interactive term on exchange rate volatility is non-negative is rejected at the 10% significance level, which implies that a negative relationship between trade finance and exchange rate volatility is strengthened during the crisis. Therefore, we argue that trade finance during the Asian crisis is exasperated predominantly by an increase in exchange rate volatility. 21
The next three columns in Table 2 show the results for the recent global financial crisis (CRISIS_2). The coefficients of the interactive dummies both for TED spread (TED) and Korean spread (KSPRD) have negative signs. However, the hypothesis testing suggests that the negative relationship is strengthened during the recent crisis only between trade finance and TED spread, which is at the 5% significance level. Hence, in the case of the recent crisis, we contend that an increase in TED spread is the driving force behind the worsening of trade finance. The last three columns in Table 2 show the results when both the Asian crisis and the global crisis are considered as a whole (CRISIS_3). The coefficients of the interactive dummies both for exchange rate volatility and TED spread have negative signs, but we cannot reject the null hypothesis that the coefficient of the interactive term on either variable is non-negative at any conventional significance level. These results suggest that none of the variables we are considering have a stronger negative relationship with trade finance during the entire period of the two crises. However, these results are not totally unexpected given the earlier finding that the signs of the interactive dummy coefficients of exchange rate volatility between the two crises are opposite, and the same is observed for TED spread. [Table 2 insert here] The results thus far imply that trade finance measured by foreign trade loans during the Asian crisis was worsened mainly by an increase in exchange rate volatility, while during the recent global crisis, it worsened by an increase in TED spread. To put it differently, there is an asymmetric relationship between trade finance and the other variables across the two crisis events. Therefore, we conclude that the two crises may adversely affect Korea s trade financing 22
condition through a different channel. However, this result is not borne out in the estimation with two lags, as shown below. The second row in Table 2 exhibits the estimation results for the two-lag specification while still using foreign trade loans. Regardless of which crisis episode we consider, we are unable to reject the null that the sum of the interactive dummy coefficients is non-negative for any of the three variables. In other words, when estimating with two lags, the deterioration of none of the three variables necessarily implies the aggravation of Korea s trade finance during the crisis periods. 22 We now turn to documentary bills as a proxy for trade finance. Since the data of documentary bills are available from 1999, we can consider only the recent crisis (CRISIS_2). Under the specification with one lag, we cannot reject the null hypothesis for any variables at a conventional significance level (the first row in Table 3). On the other hand, in the three-lag specification, we can reject only the null that the coefficient of the interactive dummy on KSPRD is non-negative at the 10% significance level (the second row in Table 3). In a nutshell, when we use documentary bills as a measure for trade finance, we find that only the adverse impact of Korean Spread on trade finance is strengthened during the recent crisis, although this finding is not robust in lag specification. [Table 3 insert here] 22 One possible reason for the insignificant results in the two-lag specification might be related to the short persistence as found in the VAR estimations; the impulse responses may be too short to be captured in the specification with two lags. 23
6. Concluding remarks In this paper, we empirically assess the impacts of financial shocks on trade finance. In so doing, we use two novel measures for trade finance in Korea: foreign trade loans extended by commercial banks, and documentary bills bought by them such as LCs and D/P or D/A. Both are representative bank-intermediate instruments of trade finance in Korea. The former supplies exporters with working capital before shipping, while the latter provides them with funds after shipping, but before importers payments. We find that regardless of which measures are used for trade finance, the impulse responses of trade finance with respect to key financial variables such as TED spread, exchange rate volatility, and Korean spread, are all negative but short-lived only for 2-3 months. Given this finding, we presume that a brief deterioration of trade finance does not contribute much to the decrease in Korea s exports during the recent crisis event. 23 However, this supposition warrants a further investigation, which is left for future research. In the case of foreign trade loans as a measure for trade finance, we find that the impacts of financial shocks on trade finance become strengthened in periods of crises, though the result is sensitive to lag-specification. Related to this, one interesting finding is that there are asymmetric impacts of the financial variables on trade finance across two financial crises in Korea. The deterioration of trade finance during the Asian crisis is primarily driven by high exchange rate volatility or country risk, whereas during the recent global crisis, it is driven by the liquidity constraint in the global financial market. 23 Due to the recent crisis, Korea s exports declined by 14% in 2009, compared with the previous year. 24
Before closing, we would like to discuss why the effects on trade finance are such shortlived. One reason might be that products of trade finance are generally considered as relatively secure compared to other standard credit lines or working capital loans, which may have less acute impacts because of financial shocks. They have a short maturity, generally up to 180 days, and are highly collateralized in the sense that credits are provided against the traded goods whose value can be calculated and secured. Also, trade finance is relatively illiquid because it cannot easily be diverted for other purpose (Chauffour and Farole, 2009). Another possible answer might be related to policy interventions in the trade finance market by the Korean government during financial crises. For instance, in May 1998, the Korea Export-Import Bank was allowed to accept documentary bills already bought by commercial banks, which, in turn, helped commercial banks buy them with fewer concerns of a default risk (Song, 1999). During the recent crisis, the Financial Supervisory Service of Korea set up a help desk for addressing the difficulties facing exporters in accessing trade finance in November 2008 (Financial Supervisory Service, 2008). Also, the Korea Export-Import Bank issued $1.5 billion in dollar-denominated bonds to provide trade finance for exporters (Chauffour and Farole, 2009), and the Bank of Korea provided $10 billion of its foreign exchange reserves to supply dollars to local banks through repurchase agreements (Auboin, 2009). These policy interventions may alleviate the impacts of financial shocks on trade finance and cause them to be short-lived. 25
References Ahn, J, M. Amiti, and D. Weinstein, 2011, Trade finance and the great trade collapse, American Economic Review: Papers & Proceedings, Vol. 101, No. 3, pp. 298-302. Amiti, M. and D. Weinstein, 2011, Export and financial shocks, Quarterly Journal of Economics, Vol. 126, No. 4, pp. 1841-1877. Auboin, M., 2009, The challenges of trade financing, VOX, January 28th. Auboin, M and M. Meier-Ewert, 2003, Improving the availability of trade finance during financial crises, World Trade Organization. Bijapur, M., 2010, Does monetary policy lose effectiveness during a credit crunch? Economic Letters, Vol. 106, pp. 42-44. Chauffour, J. and T. Farole, 2009, Trade finance in crisis: market adjustment or market failure, Policy research working paper 5003, World Bank. Cheung, W., S. Fung, and S. Tsai, 2010, Global capital market interdependence and spillover effect of credit risk: evidence from the 2007-2009 global financial crisis, Applied Financial Economics, Vol. 20, Issue 1-2, pp. 85-103. Enders, W, 2004, Applied Econometric Time Series, 3rd edition, NJ: John Wiley & Sons. FImetrix, 2009, IMF-BAFT trade finance survey: a survey among banks assessing the current trade finance environment. Financial Supervisory Service, 2008, Consultation of difficulty with trade finance and its performance, Press release, December 15 th (in Korean). International Chamber of Commerce, 2009, Rethinking trade finance 2009: an ICC global survey, Document No. 470-1120 TS/WJ 31 March 09. International Monetary Fund, 2003, Trade finance in financial crises: assessment of key issues. Levchenko, A., L. Lewis, and L. Tesar, 2010, The collapse of international trade during the 2008-2009 crisis: in search of the smoking gun, IMF Economic Review, Vol. 58 (2), pp. 214-253, December. Love, I., L. Preve, and V. Sarria-Allende, 2007, Trade credit and bank credit: evidence from recent financial crises, Journal of Financial Economics, 83, pp. 453-469. 26
MK Business News, 1998, Abolition of upper limit of trade finance for exporters, April 11th (in Korean). Pesaran, H. and Y. Shin, 1998, Generalized impulse response analysis in linear multivariate models, Economics Letter, Vol. 58, pp. 17-29. Ronci, M., 2004, Trade finance and trade flows: panel data evidence from 10 crises, IMF working paper WP/04/225, IMF. Sims, C., 1980, Macroeconomics and Reality, Econometrica, Vol. 48, pp. 1-49. Sims, C., J. Stock, and M. Watson, 1990, Inference in Linear Time Series Models with some Unit Roots, Econometrica, Vol.58, pp. 113-144. Song, S., 1999, Vitalization of trade finance for Korean firms in the era of IMF financial aids, Journal of Establishment Information, Vol, March, pp. 93-104 (in Korean). World Trade Organization, 2008, Lamy warns trade finance situation deteriorating, News items, General Council, November 12th. Yiu, M., W. Ho, and D. Choi, 2010, Dynamic correlation analysis of financial contagion in Asian markets in global financial turmoil, Applied Financial Economics, Vol. 20, pp. 345-354. 27
Table 1. Overview of Trade Finance in Korea Types Bank-intermediate instruments Instruments Working capital loans: foreign trade loans Letter of Credits (LCs) Documents against Payment (D/P) and Documents against Acceptance (D/A) Other financial institute- Export credit guarantee/insurance intermediate instruments Direct extension of credit Advance payment or later payment (open account) between traders: trade credit 28
Table 2. Interactive dummy coefficients or their sums in the VAR with foreign trade loans Lags CRISIS_1 CRISIS_2 CRISIS_3 FXSTD TED KSPRD FXSTD TED KSPRD FXSTD TED KSPRD One lag Two lags -016* (011) <070> -002 (012) <0.136> 049 (049) <0.846> 070 (052) <0.826> 011 (016) <0.758> 021 (017) <0.792> 018 (012) <0.934> -003 (014) <0.148> -066** (035) <032> -0.104 (037) <0.995> -006 (017) <0.361> -033 (018) <0.928> -004 (018) <003> -016 (021) <0.541> -045 (038) <0.119> -084 (042) <0.952> 012 (023) <0.701> -010 (026) <0.309> Notes: The number of observations is 180. The dependent variable is the first-difference of foreign trade loans (TFBOK_EX). CRISIS_1, CRISIS_2, and CRISIS_3 indicate the Asian crisis, the recent global crisis, and both crisis episodes, respectively. We report the interactive dummy coefficients in the one-lag specification or the sums of the interactive dummy coefficients in the two-lag specification. The number in ( ) is the standard error. The number in < > is the p-value of one-sided t test for the one-lag specification or one-sided Wald test for the two-lag specification with the null hypothesis that the interactive dummy coefficient or the sum of the interactive dummy coefficients is non-negative. ** and * indicate rejection of the null at the 5% and 10% significance level, respectively. 29