Exchange Rate Exposure Elasticity of Korean Companies: Pre- and Post-Economic Crisis Analysis

Transcription

1 Exchange Rate Exposure Elastcy of Korean Companes: Pre- and Post-Economc Crss Analyss Byung-Joo Lee Department of Economcs Unversy of Notre Dame Notre Dame, IN 46556, USA Phone: Fax: e-mal: JEL Classfcaton Number: F31, G1, C2 Key Words: Korean economc crss, Exchange rate exposure, Company valuaton. Abstract: Ths paper examnes the relatonshp between exchange rate movement and ndvdual company s valuaton, or exchange rate exposure, n a small open developng economy, specfcally South Korea. Ths paper found that about 2% of the Korean companes stock performance are statstcally sgnfcantly mpacted by the exchange rate fluctuatons. Ths paper also found that there were structural changes n the relatonshp between the exchange rate return and the stock returns before and after the economc crss. One of the nterestng fndngs of ths paper s that the currency deprecaton has sgnfcantly dfferent mpact on the larger companes and on the smaller companes. Before the economc crss, the currency deprecaton mproved the smallest 5 companes competveness n the nternatonal markets, thus mprovng ther stock returns. After the crss, the stock returns of the smallest 5 companes are negatvely correlated to the currency deprecaton. Ths s largely due to the short-term foregn debt oblgatons, recessonary economc condons, and government s polcy heavly favorng larger chaebol companes durng the economc crss perod. 1

2 1. Introducton It s wdely beleved that the exchange rate movement s closely related to a country s macroeconomc performance. The exchange rate s one of the most mportant prces n the country. It s the prce of one country s currency n terms of another, and, as such, converts prces denomnated n one currency nto prces denomnated n another currency. Changes n the exchange rate should therefore have a fundamental effect on the country s macroeconomc polcy varables. At the mcroeconomc level, there can also be a sgnfcant mpact on the performance of companes nvolved n nternatonal actves. For example, currency deprecaton wll ncrease the competveness of a country s export ndustry by reducng the cost of exports whle currency apprecaton wll help mport ndustry by reducng the mport prces. There have been several studes examnng how ths macroeconomc effect s transmted nto the valuaton of ndvdual company at the mcroeconomc level, especally for the multnatonal companes. Recent studes, Jorn (199), Amhud (1994), Bodnar and Gentry (1993), Bartov, Bodnar and Kaul (1996), Bodnar and Wong (2) and Gao (2) have emprcally examned U.S. multnatonal companes exchange rate exposures,.e., the relatonshp between changes n the value of the U.S. dollar and the changes n the value of the companes measured by stock prces. Huang and Stoll (1996) used hgh frequency data to examne the relatonshp between exchange rate and company s stock performance. However, most of these studes found very ltle evdence between the exchange rate movement and the company s valuaton. Bartov and Bodnar (1994) extended the prevous study by examnng not only the contemporaneous changes n the exchange rate but also the lagged change of the exchange rate. They acheved the modestly lmed success to lnk the exchange rate movement and the company s valuaton by usng the lagged exchange rate movement. Bondar and Wong (2) also acheved a modest success on explanng the exchange rate movement and company s market valuaton usng dfferent measures of market portfolo and analyzng sze effect on the exchange rate exposure. Gao (2) reexamned the foregn exchange rate exposure for U.S. multnatonal companes and found the tme varably of exchange rate exposure by ntroducng companes foregn sale and foregn producton nformaton. In contrast to 1

3 these results, Grossman and Levnsohn (1989) found statstcal sgnfcances between the exchange rates and stock returns usng sx mport competng ndustry level data. There are varous explanatons about why s hard to fnd statstcally sgnfcant evdence between exchange rate movement and the company s valuaton, even though economc theory clearly ndcates the relatonshp. Frst, multnatonal companes often reduce ther foregn exchange exposure by dversfyng ther operatons across dfferent countres. Ths s accomplshed by optmally locatng producton facles, sources of nputs, and markets for ther products. Thus, the valuaton effect of the exchange rate movement may be offset over tme as companes undertake the operatonal changes n response to exchange rate movement. Second, companes engage actve rsk management and hedge foregn exchange exposure through fnancal nstruments that offset adverse valuaton effects of exchange rate movement. Thrd, the emprcal estmates of the effect of the exchange rate change do not capture true effect of the exchange rate change. The exchange rate exposure s tme-varyng whle the majory of exstng study dd not take ths nto consderaton. Lev (1994), Amhud (1994) and Jorn (199, 1991) partally confrmed the tme-varyng nature of the exchange rate exposure, and Gao (2) successfully estmated the tme-varyng parameter of exchange rate exposure usng detaled company level data for large U.S. multnatonals. In short, whle a company s specfc economc actvy, when examned n solaton, s clearly affected by exchange rate movement, for a company as a whole, s foregn exchange rate exposure may be small or non-exstence due to the complcatng forces cancelng wh each other. Ths paper examnes the relatonshp between exchange rate movement and ndvdual company s valuaton, or exchange rate exposure, n a small open developng economy, specfcally South Korea. Ths measure s called the exchange rate exposure elastcy, and wll serve as an mportant measure to each company s responsveness to the foregn exchange volatly. One of the reasons that many of the prevous studes fal to fnd a sgnfcant connecton between exchange rate movement and a company s valuaton s that the majory of the prevous studes focus on large U.S. multnatonal companes that typcally have a relatvely small share of foregn operatons compare to the entre company s operaton. In addon, ther operatons (producton facles and 2

4 markets) are well dversfed around the world, whch provdes natural hedgng nstrument, physcally and fnancally. Bodnar and Gentry (1993) estmated exchange rate exposure n three countres, Canada, Japan and U.S. Khoo (1994) studes the Australan mnng companes, tradonally thought to be very sensve to the exchange rate movements. Other than these two studes, there are vrtually no studes examnng the companes n the small open economy that s more sensve to the exchange rate fluctuatons than the large US multnatonal companes. The next secton brefly summarzes the characterstcs of the Korean economy and the recent experences of the Korean economc crss and the more wdespread South Eastern Asan economc crss. Secton 3 presents the analytcal model and the data set. Secton 4 s an emprcal analyss of the exchange rates and the ndvdual company relatonshp n Korean economy durng the recent economc crss. Concluson and further studes are dscussed n secton Korean Economc Crss Korea s heavly dependent on the nternatonal trade for her economc survval snce the early stages of the economc development from the 196s. Therefore, s que natural that most of Korean companes have a large share of foregn operatons compared to large U.S. multnatonal companes. The foregn sales rato for the typcal large U.S. multnatonal company s about 25-4% of total sales whle typcal medum sze Korean company s foregn sales rato to the total sales s well above 5%. In addon to the sgnfcant shares of foregn sales to total sales, Korean companes heavly mport crcal ntermedate goods for fnal goods producton. Also, the majory of the Korean companes do not have sophstcated foregn currency hedgng operatons lke many of the multnatonal US companes have. Thus, Korean companes n a small open economc structure would be expected to show more dramatc effect from exchange rate movement than large U.S. multnatonal companes. In addon, many Korean companes, ncludng the large companes, so-called chaebol companes, and small ones, have sgnfcant amount of foregn debts denomnated n varous foregn currences. Ths s largely due to the growth orented expansonary polcy supported by the mplc 3

5 government guarantee of the foregn debt. Ths s the often-ced moral hazard problem, whch was one of the man causes of the recent economc crss. Therefore, the valuaton of Korean companes s very sensve to the foregn exchange rate fluctuatons. In the wake of the recent Korean economc crss startng n the late 1997, numerous companes n Korea went bankrupt as a result of the currency and bankng crses. In the early stage of the Korean economc crss, large chaebol companes lke Hanbo Steel and Ka Motors, once vrtually guaranteed by the government (too-bg-tofal), sought protecton from credors after banks refused to extend ther short-term debts. A large share of those debts s denomnated n foregn currences, and companes were not able to repay short-term debts due to the foregn currency shortages. Durng the economc crss perods, Korean companes experenced severe lqudy problems due to the large foregn debt-servcng oblgaton, thus crpplng ther normal economc operatons. Even recently, there were several notable falures and severe fnancal crss of large chaebol companes n Korea such as Hyunda Electroncs and Daewoo Motors. There have been numerous studes on the causes and consequences of the Korean and other Asan economc crss. Among others, Corsett, G., P. Pesent and N. Roubn (1998), Radelet, S. and J. Sachs (1998), and Krugman, P. (1998) analyzed the causes of the Asan economc crss n general, mostly focusng on the macroeconomc condons. Cho, D. (1999), Hahn, J. and F. Mshkn (2), and H. Root (2) nvestgated the specfc problems of South Korean economy. Km, K. and B. Lee (2) analyzed the Korean economc development and the recent economc crss focusng on the government role of the economy. However, the majory of studes focused on the macroeconomc aspects of the economc crss. Ths paper focuses on the mcroeconomc level mpact of the economc crss, more specfcally, the mpact of the exchange rate movement on the ndvdual company. It s very mportant to understand the relatonshp between the exchange rate movement and company s valuaton. Ths relatonshp wll tell whether an ndvdual company has a proper rsk dversfcaton through foregn currency hedgng operatons. A bref summary of the Korean macroeconomc condons wll help understand the mcroeconomc analyss of the exchange rate and the stock market ndex. The steady growth of the Korean economy durng 199s came to a screechng halt n

6 The real GDP has grown over 5% throughout early to mddle 199s but shrank to 5.8% n The unemployment rate ncreased from the low 2% durng the mddle 199s to 7% n Foregn debt has ncreased to $158 bllons at the end of The followng fgure wll gve you a bref dea how exchange rates and Korean stock market ndex have evolved throughout the 199s of tranqul perods and turmol perods. Fgure 1: Rate of Return for the Monthly Nomnal Exchange Rate (Won/Dollar) and the Korean Stock Market Index (KOSPI) begnnng January 199 to August KOSPRET EXRT KOSPRET = ln( KOSPI t ) ln( KOSPI t 1 ), EXRT = ln( ExchangeRatet ) ln( ExchangeRatet 1 ) From Fgure 1, s clear that the nomnal exchange rates were stable from early 199s to the mddle of These are the perods of the managed floatng exchange regme of the Korean Won. The market average exchange rate (MAR) system was adopted n March 199. Under the MAR regme, the Korean exchange rate has been determned prmarly by the market forces, wh the occasonal governmental nterventon throughout the early to the mddle 199s. When the economc crss h the South Eastern Asan regon durng the early and the mddle of 1997, Korean Won has been deprecated sharply, and the MAR system could not be mantaned. In December 1997, the free floatng exchange rate system was adopted to determne the nomnal exchange rate. Snce then, the Korean Won deprecated from around W8/$ to W2,/$ at the end of 1997, and gradually stablzed around W1,2-1,4/$ n 1998 and after. 5

7 Throughout the 199s, Korean stock market ndex was much more volatle than the nomnal exchange rates. The volatly has ncreased dramatcally after the economc crss. Durng ths perod, the returns of the stock market ndex and the exchange rate (KOSPRET and EXRT) are negatvely correlated wh correlaton coeffcent of.33. When we break 199s nto two sub-perods, tranqul perod and turmol (crss) perod, where July 1997 s the begnnng of the economc crss, the tranqul perod stll has the negatve correlaton coeffcent of.22, and the negatve correlaton coeffcent between these two varables even ncreases (n absolute value) to.39 durng the crss perod. We need more through macroeconomc analyss to draw any statstcal concluson between these two varables. However, ths smple correlaton shows that the Korean economy as a whole dd not take advantage of the export prce advantage offered by the currency deprecaton. Ths paper focuses on the mcroeconomc mpact of the exchange rates on the ndvdual company. It s also very nterestng to compare the responses of the ndvdual company s valuaton to the foregn exchange movement before and after the economc crss. One of the several hypotheses to test n ths paper s that Korean companes market valuaton to the exchange rate exposure behaves dfferently before and after the economc crss. Before the economc crss, export-orented Korean companes responded posvely (ncrease of the company s valuaton) to the currency deprecaton and negatvely (decrease of company s valuaton) to the currency apprecaton. Ths s the expected consequence of the exchange rate movement under the normal economc condons. However, when economc crss hs the Korean economy, and the domestc currency deprecates, company s foregn denomnated debt overburdens s normal economc operaton and cash flow, thus crpplng company s economc actvy, and s valuaton plummets. Ths s also largely due to the eroson of nvestors confdence on Korean companes durng the economc crss and recesson perod. In ths scenaro, the standard economc theory fals to explan company specfc response to the exchange rate fluctuatons. Therefore, ths paper wll nvestgate the tme varyng nature of the foregn exchange rate exposure elastcy before and after the Korean economc crss. The tme varyng nature of foregn exchange rate exposure was a key pont n Gao s (2) study of the U.S. multnatonals. Ths paper wll help dentfy an ndvdual company s 6

8 exposure elastcy and gve approprate strategc recommendatons about the foregn operatons dependng on the company characterstcs. Next secton develops the econometrc models to estmate the exchange rate exposure, and the detaled descrpton of the data set used n ths study. 3. Econometrc Model of Exchange Rate Exposure Elastcy The basc econometrc model s based on the market model of the Capal Asset Prcng Model (CAPM) wh exchange rate change beng addonal explanatory varable. Indvdual company s excess stock return s a functon of the excess market return and the exchange rate return. The estmaton model s: (1) r rft = α + β ( rmt rft ) + γ et + ε In ths model, r s company s stock rate of return at tme t, r mt s the market rate of return at tme t, r ft s the rsk-free nterest rate, and e t s the rate of change of exchange rate (Korean Won/US $). The rate of returns for company, market and the exchange rate are measured as the natural log dfference between tme t and (t-1) ( ln( P ) ln( P ) r, r ln( KOSPI ) ln( KOSPI ), and ln( ExR ) ln( ExR ) =, t 1 mt = t t 1 e ). t = t t 1 The rsk-free rate, r ft, s the 91 day drect depos rate. ( r ft ) stock return for company and ( rmt r ft ) s the excess market return. β measures r s an excess rate of company s systematc market rsk, and γ s company s exchange rate exposure elastcy. In a typcal market model, the unsystematc component s u = γ e + ε. In ths model, we would lke to separate the exchange rate effect component from the total unsystematc components of the fluctuaton. Snce the exchange rate s measured as Korean Won per US dollar, the ncrease of e t represents the deprecaton of Korean Won aganst US dollar. Under the normal macro economc condons, when Korean Won deprecates, Korean export prce becomes cheaper and more competve n the nternatonal market. The currency deprecaton wll help export companes sales and ther cash flow. Ths wll ncrease the (dscounted) present value of the company and t 7

9 eventually lead to the ncrease of the company s stock prce. Therefore, we expect the estmates of γ to be posve under the normal economc condons. However, after the recent Asan economc crss, company s foregn debt burden may have overwhelmed any posve cash flow effect of the currency deprecaton. Therefore, we expect that the economc crss has the negatve mpact on the γ. Ths wll help test the tme-varyng nature of the exchange rate exposure before and after the economc crss. The econometrc model s estmated by the two-stage estmaton process suggested by Levnsohn and MacKe-Mason (199) due to the bas problem of γ estmaton from equaton (1). The frst stage estmates the market model whout the exchange rate changes. Then, we can estmate the unsystematc or company specfc fluctuaton from the market model. The exchange rate exposure elastcy wll be estmated n the second stage usng the resduals (unsystematc fluctuatons) from the frst stage regresson as: (2) uˆ = ( r r ) αˆ ˆ β ( r r ) ft mt ft In the second stage, the unsystematc rsk s regressed on the several lagged values of the return of the exchange rate changes. The second stage regresson wh q lagged values of the return of the exchange rate s as follows: (3) uˆ = γ + γ jet j+ 1 + δ q j= 1 Ths regresson provdes the unbased estmates of elastcy of the exchange rate exposure, γ j. The data comes from the DataStream. The varables that we used n ths analyss are: ndvdual company stock prce ( P ), Korean stock market ndex (KOSPI t ), total market valuaton for each company measured as the stock prce multpled by the total number of the outstandng shares, and the offcal nomnal exchange rate measured as the Korean Won per US dollar (ExR t ). Data frequency s monthly and data perod starts from the begnnng of 199 when Korea opened her economy n a lmed sense, and ends at 8

10 August 2. Korean economc crss began n early September However, the general economc condons n the South Eastern Asa began to deterorate n the early and the mddle of Therefore, the data s dvded nto two perods, before crss and after crss at the begnnng of July There are 722 companes lsted n the Korean Stock Exchange at the end of August 2. However, n order to compare the before and after economc crss effect of the exchange rate changes properly, we restrct our sample to companes wh the complete prce nformaton for the entre perod begnnng January 199 to August 2. Ths reduces our sample to 56 companes. The lst of companes ncludes large chaebol companes (such as Samsung Electroncs, Hyunda Motor) as well as small to medum companes. Ths lst also ncludes all branches of Korean ndustres from bankng, textles to petrochemcals and pharmaceutcals. Exchange rate exposure wll be dfferent from one company to another dependng on s characterstcs. Therefore, I would lke to examne how certan company specfc characterstcs such as company sze, external (foregn) debt sze, profably, export rato to total sales wll affect the exchange rate exposure. In the small open developng economy, the bgger company tends to have more foregn operatons than a smaller company does. We expect that the larger company wll be more sensve to the exchange rate fluctuatons. In the South Eastern Asan developng countres, s true that the larger companes have accumulated more foregn debts for the rapd expanson under the mplc government loan guarantee. They are the ones that have the so-called moral hazard problem. However, as we have seen n the prevous lerature, large multnatonal companes have well dversfed ther foregn exchange exposure physcally and fnancally. Therefore, was very hard to fnd the emprcal evdence of the foregn exchange exposure. It wll be nterestng to see whether the larger companes n the small open developng countres have the smlar sophstcaton of the large US multnatonals to dversfy foregn exchange rsks. 4. Emprcal Estmates of the Exchange Rate Exposure Elastcy Ths secton reports detaled emprcal fndngs of the exchange rate exposure elastcy of Korean companes before and after the economc crss. Frst, we estmated 9

11 the market model for two sub perods to see f there were any statstcally sgnfcant structural changes before and after the Korean economc crss. Ths model s estmated by usng the dummy varable to dstngush two sub perods. (4) r rft = α + β ( rmt rft ) + α1 d + β1 d ( rmt rft ) + ε where the dummy varable d equals to one for crss perod (from July 1997 to August 2), and zero otherwse. Ths model wll help to dstngush the structural changes of the market model and provdes the statstcal tools to test the hypothess of the structural changes. Equaton (4) s estmated by the smple least squares estmaton. Whe s (198) heteroskedastcy consstent covarance estmaton method s used to estmate the consstent covarance matrx of the parameters. The estmaton results are reported n Table 1.1. Table 1.1: Estmates of the Market Model wh Crss Dummy Parameters Mean Standard Statstcal Sgnfcance Devaton P5% P1% N5% N1% α o β α β Mean estmates are the sample average of 56 companes estmates. For example, α ˆ 56 1 = ˆ 56 α = = 1.9. The standard devaton s the standard devaton of 56 estmates. The numbers n the statstcal sgnfcance columns (P5%, P1%, N5%, N1%) are the number of estmates that are statstcally sgnfcant at 5% and 1% for posve and negatve values. Ths table reports that the average market beta ( β ) of 56 companes s.825 and s standard devaton s Also, reports that 491 out of 56 company s betas are statstcally sgnfcant at 9%. For dummy varable parameters, we can see that none of the ntercept dummy parameters ( α 1 ) are posvely statstcally sgnfcant whle 24 out of 56 companes beta dummes are statstcally sgnfcant at 5%. These results show that the market model was generally stable for the majory (more than 95%) of companes before and after the economc crss. The market stably hypothess s tested 1

12 as follows. The null hypothess s: H α = β. For all 56 companes, 27 : 1 1 = companes rejected ths hypothess usng F-test at 5% sgnfcance level. Ths s slghtly greater than 5% of the total companes, so we can reasonably consder ths as the random varaton. Therefore, to test the exchange rate exposure elastcy, we used the stable market model for the entre perods and obtaned the resdual from the sngle market model. The estmaton result s reported n Table 1.2. The resdual from the stable market model s used n the second stage regresson to estmate the exchange rate exposure elastcy. Table 1.2: Estmates of the Market Model whout Crss Dummy Parameters Mean Standard Statstcal Sgnfcance Devaton P5% P1% N5% N1% α o β In the second stage regresson, varous specfcatons of the lagged values of the return rate of the exchange rate are consdered. The same dummy varable s used to dstngush the effect of the economc crss. The second stage regresson model s: q (5) uˆ = γ + γ j et j+ 1 + γ 1 d + γ 1 j d et j+ 1 + δ j= 1 q j= 1 Ths regresson s also estmated usng the Whe s (198) heteroskedastcy consstent covarance estmaton method. For varous specfcatons of the lag length, the estmaton results are qualatvely smlar and we chose the lag length 4 accordng to the Akake nformaton crera. We estmate equaton (5) for two separate models. The frst model does not nclude the crss dummy and the second model ncludes the crss dummy varable. These two estmaton results show nterestng comparson of the structural changes of the before and after economc crss. The emprcal results are reported n Table 2.1 and Table 2.2. Table 2.1: Exchange Rate Exposure Elastcy whout Crss Dummy Parameters Mean Standard Statstcal Sgnfcance Devaton P5% P1% N5% N1% 11

13 γ γ γ γ γ γ Table 2.1 shows that the estmates of the exchange rate exposure elastcy throughout the 199s are negatve. The contemporaneous, the frst and the second lagged elastcy estmates are all negatves ( γ 1 =. 5585, γ 2 = and γ 3 =. 826 ). There are 22 and 18 companes statstcally sgnfcant at 1% that have the negatve contemporaneous and the frst lagged elastces. The second lagged elastcy does not have any clear trend. These results reveal that the Korean companes as a whole dd not take advantage of the currency deprecaton for the export competon throughout the 199s. Ths fndng s consstent wh the macroeconomc evdence that we have seen n Secton 2. Throughout the 199s, the returns of the stock market ndex and the exchange rate are negatvely correlated wh correlaton coeffcent of.33. However, f we ntroduce the structural breakdown for the dramatc changes of the macroeconomc condons, there are very dfferent and nterestng results. Table 2.2: Exchange Rate Exposure Elastcy wh Crss Dummy Parameters Mean Standard Statstcal Sgnfcance Devaton P5% P1% N5% N1% γ γ γ γ γ γ γ γ γ γ

14 γ γ Table 2.2 reveals very nterestng emprcal evdences dfferent from those of Table2.1. Frst of all, the contemporaneous (lag zero) and the lag one exchange rate exposure elastces are posve ( γ = and γ = ) before the economc crss and ther correspondng dummy parameters measurng the effect of the economc crss are negatve ( γ 11 = and γ 12 = ). Ths explans the effect of the exchange rate changes on the company valuaton n the stock market. Before the economc crss, the deprecaton (posve return of the exchange rate changes) ncreases the company valuatons n general, thus the posve parameters. However, after the economc crss, the deprecaton affects the company valuaton negatvely. Ths s largely explaned by the heavy debt burden denomnated by the foregn currences. On the eve of the economc crss, large Korean company s debt equy rato was about 5%, far greater than the nternatonal norm. Even though the statstcal sgnfcance of the parameters s not very mpressve, s stll a very mportant fndng that we verfed our conjecture that the exchange rate elastcy exposure has changed s sgn after the economc crss. For the frst lagged elastcy estmates, we have 117 posvely statstcally sgnfcant at 1% sgnfcance level whle 143 companes have negatvely statstcally sgnfcant dummy parameter estmates after the economc crss. The statstcally sgnfcant parameters represent only less than 3% of the entre 56 companes. But s well documented that s not unusual for falure of fndng any meanngful statstcal estmates for the exchange rate elastces due to the complcated nature of company operatons and sophstcated hedgng strateges to dversfy exchange rate rsks. To examne more detals about the company specfc exchange rate exposure elastcy, the followng four Fgures show the emprcal dstrbutons of the current and the frst lag parameters before and the after the economc crss. 13

15 Fgure 2.1: Emprcal Dstrbuton of γ 1 ( γ 1 =Seres B22) Seres: B22 Sample Observatons 56 Mean Medan.241 Maxmum Mnmum Std. Dev Skewness Kurtoss Jarque-Bera Probably.317 Fgure 2.2: Emprcal Dstrbuton of γ 2 ( γ 2 =Seres B23) Seres: B23 Sample Observatons 56 Mean Medan Maxmum Mnmum Std. Dev Skewness.5498 Kurtoss Jarque-Bera Probably. 14

16 Fgure 2.3: Emprcal Dstrbuton of γ 11 ( γ 11 =Seres B32) Seres: B32 Sample Observatons 56 Mean Medan Maxmum Mnmum Std. Dev Skewness Kurtoss Jarque-Bera Probably.28 Fgure 2.4: Emprcal Dstrbuton of γ 12 ( γ 12 =Seres B33) Seres: B33 Sample Observatons 56 Mean Medan Maxmum Mnmum Std. Dev Skewness Kurtoss Jarque-Bera Probably. 15

17 Fgures 2.1 and 2.2 show the emprcal dstrbutons of the current and the frst lagged parameters for the exchange rate exposures before economc crss. Even though there are only 38 and 117 statstcally sgnfcant parameters, respectvely, at 1% sgnfcance level, they have posve means ( γ = and γ = ) and posve skewness (.24 and.541). Fgures 2.3 and 2.4 are the emprcal dstrbutons of the current and the frst lagged dummy nteracton parameters for the exchange rate exposures after the economc crss. They have negatve means ( γ 11 = and γ 12 = ), and negatve skewness ( and.4668). From Table 2, we can also fnd that the current and the frst lagged exchange rate changes affect company valuatons most. In fact, the second and the thrd lagged parameters have the oppose sgns from the current and the frst lagged parameters. These sgns turn from posve to negatve before the crss whle they turn from negatve to posve after the crss. The deprecaton of the Korean currency aganst the US dollar affected companes negatvely even before the economc crss. Ths explans the fragle nature of the Korean companes fnancal structure. Many Korean companes have operated under the large amount of foregn debt and most of the debts were shortterm debts. Even before the economc crss, the benefs of the currency deprecaton are short lved. Whle the currency deprecaton ncreased company s competveness for the short-term perod, after few months later, the deprecated currency quckly turned nto the negatve factors as ncreased the short-term debt servcng oblgatons. In fact, they are fnancally strapped by any foregn currency fluctuatons. To test the structural changes of the relatonshp between the unsystematc rsk and the rate of the exchange rate before and after the economc crss, F-test was performed for the null hypothess H γ = γ = γ = γ = γ = γ. We rejected : = the null hypothess 127 out of 56 companes at 5% (one-sded) sgnfcance level, and 181 companes at 1% sgnfcance level. Ths represents about 25% to 35% of all sample companes. Whle ths result does not seem to be very mpressve, ths fndng produces much better results than any of the exstng studes regardng the exchange rate exposure elastcy estmaton. One of the reasons for ths success may be that the majory of the Korean companes, large and small, rely heavly on the nternatonal trade, and they are very sensve to the foregn currency fluctuatons. 16

18 Whle Table 2 provdes the summary statstcs for the parameter estmates of the exchange rate exposure elastcy of the Korean companes, s very hard to fnd any specfc characterstcs of ndvdual company. In order to nvestgate more detaled characterstcs of the exchange rate exposure elastcy, two sub samples are constructed: the top 5 largest companes and the bottom 5 companes accordng to the market valuaton at the end of December December 1996 valuaton was used to dstngush the top 5 and bottom 5 companes because the next several months have brought the dramatc changes of the economc condons throughout the South Eastern Asa. Before 1997, was stll stable and tranqul perod for the regon. If there are any dfferences to the exchange rate changes between large companes and small companes, these two sub samples wll contrast the dfferences. The followng fgures show very nterestng results for the estmates of these two sub samples. Fgures 3.1 to 3.4 show the emprcal dstrbutons of the exchange rate exposure elastces for the top 5 largest companes. From Fgures 3.1 and 3.2, s very clear that the largest companes react negatvely to the exchange rate deprecaton even before the economc crss. Ths s very dfferent from the prevous fndngs for the entre Korean companes. In general, the majory of the export orented Korean companes reacted favorably to the exchange rate deprecaton before the crss. Fgures 4.1 and 4.2 for the smallest 5 companes confrm our prevous general fndngs that the currency deprecaton wll mprove company s stock performance. However, after the economc crss, the smallest 5 companes shown n Fgures 4.3 and 4.4 react very negatvely to the deprecaton whle the largest companes reacton to the deprecaton n Fgures 3.3 and 3.4 s mldly posve. These are very sgnfcant fndngs that have never been analyzed before. Before the economc crss, the smallest companes wh the mnmal debts took the full advantage of the currency deprecaton to be more competve n the nternatonal markets, the largest companes, mostly chaebols, were not able to take advantage of the favorable exchange rate changes due to the heavy foregn debts. Ther foregn debt servcng oblgatons overwhelmed the competve advantages offered by the currency deprecaton. However, after the economc crss, whle most of the smallest companes suffer heavly due to the general recesson and the domestc economc downturn, the largest companes took advantage of ths economc downturn. 17

19 There s no clear evdence why the largest companes took advantage of the currency deprecaton whle most of the small companes suffer because of the recesson. One possble answer to ths puzzle s that the government polcy after the economc crss turned favorably to the largest companes whle neglectng the majory of the small companes. After the economc crss, n early 1998, the newly elected Korean government took the natve to reform chaebol orented ndustry structure, so-called the Bg Deals polcy. Ths s the government-controlled ndustry restructurng polcy, swappng assets of dfferent companes of dfferent ndustres among chaebol companes. Ths polcy s desgned to elmnate unnecessary competons among smlar companes n the same ndustry, and to strengthen nternatonal competveness of the chaebol companes. However, as we have seen from our emprcal evdence, the larger chaebol companes have benefed from the fall of the Korean economy whle the smaller companes are vctms of the economc crss. Government polcy needs to focus more on the smaller companes to overcome economc hardshp. 18

20 Fgure 3.1: Emprcal Dstrbuton of the Contemporaneous Elastcy ( γ 1 ) for Top 5 Companes 12 1 Seres: B22 Sample Observatons Mean Medan Maxmum Mnmum Std. Dev Skewness Kurtoss Jarque-Bera Probably Fgure 3.2: Emprcal Dstrbuton of the Frst Lagged Elastcy ( γ 2 ) for Top 5 Companes Seres: B23 Sample Observatons 5 Mean Medan Maxmum Mnmum Std. Dev Skewness Kurtoss Jarque-Bera Probably.1 Fgure 3.3: Emprcal Dstrbuton of the Contemporaneous Dummy Elastcy for Top 5 Companes Seres: B32 Sample Observatons 5 Mean.1863 Medan.1885 Maxmum Mnmum Std. Dev Skewness Kurtoss Jarque-Bera Probably Fgure 3.4: Emprcal Dstrbuton of the Frst Lagged Dummy Elastcy for Top 5 Companes Seres: B33 Sample Observatons 5 Mean Medan.3422 Maxmum Mnmum Std. Dev Skewness Kurtoss Jarque-Bera Probably. 19

21 Fgure 4.1: Emprcal Dstrbuton of the Contemporaneous Elastcy ( γ 1 ) for Bottom 5 Companes 1 8 Seres: B22 Sample Observatons Mean Medan.464 Maxmum Mnmum Std. Dev Skewness Kurtoss Jarque-Bera Probably Fgure 4.2: Emprcal Dstrbuton of the Frst Lagged Elastcy ( γ 2 ) for Bottom 5 Companes 1 8 Seres: B23 Sample Observatons Mean Medan Maxmum 6.24 Mnmum Std. Dev Skewness Kurtoss Jarque-Bera Probably Fgure 4.3: Emprcal Dstrbuton of the Contemporaneous Dummy Elastcy for Bottom 5 Companes Seres: B32 Sample Observatons 5 Mean Medan Maxmum Mnmum Std. Dev Skewness Kurtoss Jarque-Bera Probably Fgure 4.4: Emprcal Dstrbuton of the Frst Lagged Dummy Elastcy for Bottom 5 Companes 1 8 Seres: B33 Sample Observatons Mean Medan Maxmum Mnmum Std. Dev Skewness Kurtoss Jarque-Bera Probably

22 5. Concluson and Further Research Ths paper analyzed the mpact of the mcroeconomc structure of the Korean economc crss. To be more precse, ths paper nvestgated the effect of the exchange rates on the ndvdual company stock performances durng the economc crss. For large US multnatonal companes, the currency fluctuaton dd not cause any statstcally sgnfcant mpact on the stock prces thanks to the sophstcated fnancal and geographcal dversfcatons. Ths paper focuses on the companes n the small open economy, South Korea, that are much more hghly exposed to the foregn currency fluctuatons than US companes are. Ths paper found that about 2% of the Korean companes stock performance are statstcally sgnfcantly mpacted by the exchange rate fluctuatons. Ths paper found that there were structural changes n the relatonshp between the exchange rate return and the stock returns before and after the economc crss. Before the economc crss, currency deprecaton generally helped company stock performances, but after the crss, ths relatonshp has been sgnfcantly weakened, and the currency deprecaton actually hurt the majory of the Korean companes. One of the nterestng fndngs of ths paper s that the currency deprecaton has a sgnfcantly dfferent mpact on the larger companes and on the smaller companes. For the top 5 largest companes, the currency deprecaton hurt ther stock performances before the crss but helped them after the crss. Currency deprecaton affected the smallest 5 companes exactly the oppose way of the largest 5 companes. Before the crss, the currency deprecaton mproved the smallest 5 companes competveness n the nternatonal markets, thus mprovng ther stock returns. After the crss, the stock returns of the smallest 5 companes are negatvely correlated to the currency deprecaton. Ths s largely due to the short-term foregn debt oblgatons and the general economc condons of the South Korea durng the crss perod. Ths paper nvestgated the mcroeconomc relatonshps between the stock performance and the currency fluctuatons focusng on the ndvdual company. We have found margnally statstcally sgnfcant relatonshp between these two varables before and after the economc crss. However, from these results, s very hard to draw any 21

23 ndustry specfc conclusons. Naturally, there are certan ndustres (Bankng, Semconductor ndustres) that are more susceptble to the currency fluctuatons than the other (Agrcultural, Fsheres ndustres). Therefore, s the logcal next step that we extend our current research nto the ndustry specfc relatonshps. I would lke to extend the current research usng the ndustry ndex rather than ndvdual company stock prces. 22

24 References Amhud, Y. (1994), Exchange rates and the valuaton of equy shares, n Exchange rates and corporate performance, eds. By Amhud,Y and R. Levch, Irwn Professonal Publshng, New York, p Bartov, E., and G. Bodnar (1994), Frm Valuaton, Earnngs Expectatons and the Exchange rate exposure effect, Journal of Fnance, 49(5), p Bartov, E., G. Bodnar and A. Kaul (1996), Exchange rate varably and the rskness of U.S. multnatonal frms: Evdence from the breakdown of the Bretton Woods system, Journal of Fnancal Economcs, 42(1), p Bodnar, G., and W. Gentry (1993), Exchange rate exposure and ndustry characterstcs: evdence from Canada, Japan and the USA, Journal of Internatonal Money and Fnance, 12, p Bodnar, G. and M.H. Wong (2), Estmatng exchange rate exposure: Some weghty ssues, NBER workng paper Chen, N., R. Roll and S. Ross (1986), Economc forces and the stock market, Journal of Busness, 59(3), p Cho, D.C. (1999), Recoverng from the crss: Where the Korean economy stands? workng paper, Korea Development Instute. Corsett, G., P. Pesent and N. Roubn (1998), What caused the Asan currency and fnancal crss, NBER workng paper. Gao, T. (2), Exchange rate movements and the profably of US multnatonals, Journal of Internatonal Money and Fnance, 19, p Grossman, G. and J. Levnsohn (1989), Import competon and the stock market return to capal, Amercan Economc Revew, 79(5), p Hahn, J.H. and F. Mshkn (2), Causes of the Korean fnancal crss: Lessons for polcy, NBER workng paper No Huang, R. and H. Stoll (1996), Exchange rates and frms lqudy: Evdence from ADR, workng paper, Vanderblt Unversy. Jorn, P. (199), The exchange rate exposure of U.S. multnatonals, Journal of Busness, 63(3), p Jorn, P. (1991), The prcng of exchange rate rsk n the stock market, Journal of Fnancal and Quantatve Analyss, 26(3), p Khoo, A. (1994), Estmaton of foregn exchange exposure: an applcaton to mnng companes n Australa, Journal of Internatonal Money and Fnance, 13(3), p Km, K.S. and B.J. Lee (2), The 1997 fnancal crss and economc governance: The case of South Korea, workng paper, Unversy of Notre Dame Krugman, P. (1998), Asa: What went wrong? Fortune, March 2. Lev, M. (1994), Exchange rates and the valuaton of equy shares, n Exchange rates and corporate performance, eds. By Amhud,Y and R. Levch, Irwn Professonal Publshng, New York, p Levnsohn, J. and J. MacKe-Mason (199), A smple consstent estmator for dsturbance component n fnancal models, Revew of Economcs and Statstcs, 72(3), p

25 Marston, R. (1996), The effects of ndustry structure on economc exposure, NBER workng paper Radelet, S. and J. Sachs (1998), The onset of the East Asan fnancal crss, Harvard workng paper. Root, H. (2), Korea s recovery: Don t count on the government, Mlken Instute workng paper. Whe, H. (198), A Heteroskedastcy-consstent covarance matrx estmaton and a drect test for heteroskedastcy, Econometrca, 48, p