Conditions For Foreigners’ Trade Executions in The KOSPI - What Are The Determinants of Portfolio Flows of Foreign Origin? - Written By Minseok Lee Senior Honors Thesis (2011), Economics Department, Tufts University Read By Professor Heiwai Tang Professor Professor Patrick Schena Abstract This paper observes the historical price to earnings ratio (PER) and the trading data of the four major investor groups in the Korea Composite Stock Price Index (KOSPI): foreign investors, institutional investors, individual investors, and other corporations. The paper analyzes foreign investors’ role in setting market valuation-norms in the KOSPI from July 2002 to December 2010, and examines whether or not the foreign investors are most value-oriented among the four groups. This paper finds that it is unclear whether or not valuation-cause have been a meaningful and reliable metric of purchases and sales of securities for foreign investors who participate in creating the portfolio flows. However, this paper finds that foreign investors sensitively react to the anticipation of the VIX and the won-dollar exchange rate. 20110429 Minseok Lee Index I. Introduction……………………………………………………………………. 2 II. Backgrounds 1. Valuation and Trading Flow, and Value Investors…………………… 4 2. The KOSPI and Hot Money………………………………………….. 5 III. Analyses of The Residual PER’s Impact on Trading Flow …………..…….. 1. Design………………………………………………………..………. (1) Big Picture…………………………………………………. (2) Detailed Scene……………………………………..………… 2. Result……………………………………………..…………..……..... IV. Conclusion…………………………………………………………………… 1 12 20110429 Minseok Lee I. Introduction In the Korea Composite Stock Price Index (KOSPI), investors are frequently categorized into the following four groups: foreign investors, local institutional investors, individual investors, and other corporations. 1 Korea Exchange (KRX) compiles and discloses each investor group’s trading flow data at both high and low frequencies. Because there is a public access to such detailed data at high precision, each investor group’s behaviors are repeatedly tracked and characterized by journalists and financial analysts. In the KOSPI, foreign investors are generally believed to be the most sophisticated investors with the most abundant monetary supplies and best human capital. Due to the common placed dollar shortage in the won-dollar segment of the exchange (currency) market, foreign investors almost always enjoy the luxury of not having to suffer Korea’s domestic dollar shortage, and thus they often are considered as the richest investors. As foreign investors are mostly from larger economies, the scale of their capital base is much larger than that of most Korean institutions or individuals (Financial Supervisory Service, 2011). As a result, foreign investors are believed to be able to afford to execute trades with more confidence. Many contemporary investment strategy reports popularize the stereotype of foreign investors as managing the trading flow’s momentum in order to maximize the unrealized profit from the positions that they have previously made (De Long, Shleifer, 1 Except for the foreign investors, the location of trading account is in Korea. More detailed elaboration on the classification of each investor group is made in Background 2. 2 20110429 Minseok Lee Summers, and Waldmann, 1990). Consequently, foreign investors often face public criticism that their portfolio flows do not carry enough valuation-cause. While there are specific funds of foreign origin that are characterized as long term value investors, foreign portfolio flows as a whole are often depicted as if they have intention to manipulate the market with their capability and sophistication. This characterization gives the impression to the members of investment community, including regulators, that foreign investors do not take a constructive role in the KOSPI. The recent sanction on Deutsche Bank by Korea’s Prosecutors’ Office and the public criticism on the incident was one good example that showed the negative image of the portfolio flows of foreign origin (Lee and Tudor, 2011). Viewers of the incidents similar to Deutsche Bank’s case could have made a mistake of hasty generalization. In this paper, the author sought to find out whether or not foreign investors are significantly value-oriented investor group. Despite the research, the author finds that the degree of value orientation of each investor group is unclear. However, he identifies the risk appetite and the won-dollar exchange as clearly significant factors that interact with foreign investors’ portfolio flows. In the period from Jul. 2002 to Dec. 2010, foreign investors were sensitive to the VIX, and they paid more attention to the risk management than other players did. In addition, currency valuation appear not to be perceived by foreign investors as a major consideration of equity investment in Korea, unless the currency made an excessively depressive deviation from the norm in the short term. 3 20110429 Minseok Lee II. Background II. 1. Valuation and Trading Flow, and Value Investors Valuations and trading flows2 inevitably influence each other in the traded equity markets. Securities’ valuations provide a logical foundation to the value-oriented investors and encourage them to take positions. As more long (buy) positions or fewer short (sell) or no positions are made, the securities’ value appreciates or depreciates. This invites the question of which, between the valuations or the trading flows, takes a more autonomous role in determining the other factor. Value investors believe that securities have correct values that at some point should be converged upon, because “owners and buyers of common stocks are generally anxious to arrive at an intelligent idea of their value” and human nature seeks to wrap the speculative greed with “a screen of apparent logic and good sense” (Graham and Dodd, 2008).3 While value investors have historically enjoyed significant returns, those who pay close attention to the trading flows and seek to exploit the opportunities that arise from interpreting the trading flows have also been thriving, as proven by the presence of a significant number of arbitragers and quant-hedge funds.4 Many market participants sit somewhere in the middle of the two investment styles. In this paper, ‘valuations’ means justifiable values given to the securities through security analyses of various sort by market participants, and ‘trading flows’ means every disclosed transaction made by buyers and sellers of respective securities in the market. Block deals are not taken into account. 3 Variations within value investment and the reason for selecting PER as valuation metric is elaborated in the latter part of this paper. 4 Value investors call this investor group as speculators (Klarman, 2008); however, this paper defines investors as: anyone (including machines) who buys and/or sells in the market. 2 4 20110429 Minseok Lee Although different investors have various areas of focus in terms of their investment analysis, the executions of their trades exhibit only two forms: either long or short.5 Value investors seek to buy an asset at a discount and sell it at a premium, and when these intentions are combined with the fact that buyers and sellers determine the price of any traded security, it is evident that even purely value-oriented investors add and reduce momentum. Buying at bid price or selling at offer price represents the cases of ‘adding momentum to the desired direction’. Buying at offer price or selling at bid price represents the cases of ‘reducing momentum from the opposite direction’. This is done through unintentionally adding momentum to the desired direction or reducing momentum from the opposite direction in order to set the valuation level that they desire to achieve in the investment process.6 II. 2. The KOSPI and Hot Money The Korean equity market has developed in scale and sophistication as the listed companies and the capital markets have grown (Figure 3). The KOSPI comprises of many significant companies, most of which have long history and large market capitalizations. The saturation in terms of market capitalization is relatively high due to Korea’s economic development, which was dominated by large conglomerates (Figure 1 and 2). 5 In this paper, positions made by arbitragers who often seek to eliminate their risks are not discussed in detail. However, purchased-sold history of equities enables the author to successfully include the arbitrageurs either directly or indirectly. 6 A true distinction between value investor and momentum investor can only be drawn from their approach and attitude rather than from their investment executions. When the size of investment is large relative to the market depth, it becomes more difficult to differentiate value investors from a trader or a momentum investor. This is because it is easier for value investors to set the valuation norm when purchased-sold amount per total traded amount is high. Support this with analysis: each region’s various markets’ turnover versus their volatilities. 5 20110429 Minseok Lee Figure 1. KOSPI Saturation by Market Capitalization (Mar. 2011) Market Capitalization Sort Large Cap. (Million Total Market Capitalization Largest In The Group Smallest In The Group Korean Won) Average Market Capitalization Weight Number of Companies 908,683,988 139,197,873 961,174 9,086,840 87% 100 Mid Cap. 109,989,064 3,483,149 71,307* 555,500 10% 198 Small Cap. 30,562,647 487,140* 2,719 70,747 3% 432 All 1,049,235,699 139,197,873 2,719 1,437,309 100% 730 Source: KRX * This discrepancy attributes to the abnormal short-term price fluctuation during the period of analysis. ** The Sorting was done on the 4th of Mar. 2011. Figure 2. KOSPI Saturation by Market Capitalization (Mar. 2011) Category Market Capitalization (Million Korean Won) Weight 1* 139,197,873 13.3% 1~10 396,230,295 37.8% 1~50 767,705,199 73.2% 1~100 908,683,988 86.6% All 1,764,989,556 100% Source: KRX *Samsung Electronics Corporation Figure 3. KOSPI Market Capitalization (Jan. 2001 – Dec. 2010, Daily) 6 KRW 20110429 Minseok Lee 1.20E+09 1.00E+09 8.00E+08 6.00E+08 4.00E+08 2.00E+08 01.1.2 01.5.15 01.9.20 02.1.31 02.6.17 02.10.25 03.3.6 03.7.15 03.11.25 04.4.8 04.8.16 04.12.22 05.5.4 05.9.9 06.1.17 06.5.26 06.10.9 07.2.14 07.6.26 07.11.6 08.3.19 08.7.29 08.12.4 09.4.15 09.8.20 09.12.24 10.5.6 10.9.10 0.00E+00 Source: Bloomberg LP *KRW: Korean Won Major shareholding investor groups in KOSPI include foreign investors, institutional investors, individual investors, and other corporations. 7 Foreign investors include those without foreigner investment registration identification as well as ones with the identification. Institutional investors includes local mutual funds, investment banks, foundations, government pension funds, insurance companies, commercial banks, merchant banks and private equity funds whose accounts are located in South Korea. Technically, individual investors include individuals who trade with personal stock accounts; however, ‘individual investors’, as it is colloquially used, also add and reduce direct momentum to the mutual funds by initiating fund flows from and into the mutual 7 Major shareholding investor groups in some literature are foreign investors, institutional investors, individual investors, government bodies, and other corporations. The distinction among institutional investors, government bodies and other corporations has not been very strict in the KRX’s record; one or two types of sub categorical investor groups such as pension funds have been included under institutional investors or government bodies at different periods of time (Korea Exchange, 2011). To eliminate the confusion, the author merges government bodies and institution into one, and calls it institutional investors. In addition, the author’s focus on the foreign investors is partly attributable to this categorical issue in the data compilation. 7 20110429 Minseok Lee funds. Other corporations include venture capitals, and medium-size corporations that do not specialize in investment. All of those who are identified with their business registration identifications, due to their lack of institutional investor registration identifications, are categorized under other corporations. Each player within other corporations has various trading purposes that does not directly relate to the profit seeking; hence other corporations’ intentions and behaviors are often not generalized. The above-mentioned four players’ inflows and outflows during the same period sum to zero without an exception. The foreign investors hold large portions of KOSPI (Figure 4 and 5), and they create the bulk of trading flow (Figure 6), while individual investors’ turnover has historically been significantly higher than the rest (Figure 7). The total turnover has also been growing (Figure 8). Just like other global indices, the KOSPI shows high correlation with every other major stock index around the world (Figure 9). In May of 1998, the universal cap on foreign shareholding in KOSPI was removed by Korea’s Financial Supervisory Service, while very few strategically important sectors that provide public services remained exceptions.8 There is no additional tax applicable for foreigners. With very few exceptions,9 the daily limit for appreciation or depreciation of any share prices is ±15%. 8 The shareholding limits for foreigners range from 33% to 50% for companies such as: Korea Telecom, SK Telecom, LG U Plus, Korea Tobacco & Ginseng, KoGas, Korea Electricity Power Corporation, Korean Air, Asiana Air, and Some broadcasting systems. Foreigners’ purchase of the shares above the limit is possible with the use of derivative products, though foreigners hardly hit the shareholding limits in practice (Korea Exchange, 2011). 9 Periods of the initial public offering (IPO) and delisting fall into the exceptions. 8 20110429 Minseok Lee Figure 4. KOSPI shareholding by investor group (2000 – 2009, Yearly) Year (Billion Korean Won) Investor Group 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 GOV 26,714 20,824 14,645 16,127 18,153 26,060 30,227 29,713 16,767 16,441 INS 29,475 40,297 40,991 59,328 72,765 128,375 154,723 201,589 71,144 111,087 OTH 36,493 43,914 52,123 67,354 74,246 119,658 130,667 204,996 166,287 194,811 IND 37,314 57,117 57,760 70,020 74,266 120,718 126,437 207,465 155,703 275,253 FOR 56,209 93,698 93,161 142,534 173,158 260,263 262,534 308,181 165,655 289,724 Year (%) Investor Group 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 GOV 14.35 8.14 5.66 4.54 4.40 3.98 4.29 3.12 2.91 1.85 INS 15.83 15.75 15.85 16.70 17.64 19.60 21.96 21.18 12.36 12.52 OTH 19.60 17.16 20.15 18.95 18.00 18.27 18.55 21.53 28.89 21.96 IND 20.04 22.32 22.33 19.70 18.00 18.43 17.94 21.79 27.05 31.02 FOR 30.19 36.62 36.01 40.11 41.97 39.73 37.26 32.37 28.78 32.65 Source: KRX * GOV: Government & Public Bodies, INS: Institution, OTH: Other Corporations, IND: Individual, FOR: Foreigners. Figure 5. KOSPI shareholding by investor group (2000 – 2009, Yearly) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Foreigners Individual Other Corporations Institution Government & Public Bodies Source: KRX Figure 6. KSFINET Index (Jan. 2003 – Dec. 2010, Daily) Ticker Max Inflow (Million USD) Max Outflow (Million USD) Variance Standard Deviation KSFINET Index* 1204.02 -1175.95 54313.83 233.05 Source: Bloomberg LP 9 20110429 Minseok Lee * KSFINET Index: Daily Net Flows in South Korean equities by foreign investors in KOSPI & Korean securities dealers automated quotations (KOSDAQ). Foreign investors include one without foreigner investment registration ID as well as one with ID. **The numbers in the table above largely shows the portfolio flows. Figure 7. Each Investor Group’s Weight of Monthly Purchased-Sold Amount (Jul. 2002 – Dec. 2010, Monthly) 80 70 60 50 40 30 20 10 0 200207 200211 200303 200307 200311 200403 200407 200411 200503 200507 200511 200603 200607 200611 200703 200707 200711 200803 200807 200811 200903 200907 200911 201003 201007 201011 % Dat FOREI INSTI INDIV OTHER Source: KRX *Each investor group’s weight of monthly purchased-sold amount = (purchased amount + sold amount) / Turnover / 2 Figure 8. KOSPI Turnover (Jul. 2002 – Dec. 2010, Monthly) 2.00E+11 Korean Won 1.80E+11 1.60E+11 1.40E+11 1.20E+11 1.00E+11 8.00E+10 6.00E+10 4.00E+10 2.00E+10 Date 200207 200211 200303 200307 200311 200403 200407 200411 200503 200507 200511 200603 200607 200611 200703 200707 200711 200803 200807 200811 200903 200907 200911 201003 201007 201011 0.00E+00 Source: Bloomberg LP 10 20110429 Minseok Lee Figure 9. Correlations Among Major Equity Indices Around The World (Jan. 2002 – Dec. 2010, Daily) KOSPI S&P500 DAX FTSE CAC HSI STI NKY IBOV KOSPI 1.0000 S&P500 0.5869 1.0000 DAX 0.9133 0.8154 1.0000 FTSE 0.7611 0.9300 0.9229 1.0000 CAC 0.5412 0.9396 0.7965 0.9150 1.0000 HSI 0.9536 0.6416 0.9033 0.7585 0.5663 1.0000 STI 0.9360 0.7903 0.9691 0.8895 0.7304 0.9403 1.0000 NKY 0.4542 0.8648 0.6977 0.8175 0.9198 0.4831 0.6254 1.0000 IBOV 0.9511 0.4362 0.8167 0.6192 0.3665 0.9314 0.8633 0.2725 1.0000 Source: Bloomberg LP * Acronyms in the table indicate the following: the Standard & Poor’s 500 (S&P500), the Deutscher Aktien Index (DAX), the FTSE 100 Index (FTSE), the CAC 40 (CAC), the Hang Seng Index (HSI), the Straits Times Index (STI), the Nikkei 225 (NKY), and the Bovespa Index (IBOV). The majority of companies listed in KOSPI report their annual earnings around April. The earnings’ releases’ impact on the PER is observed in Figure 10, as most conspicuous adjustments in the PER occurred in Aprils of 2003, 2004, 2005, 2009, and 2010. One notable take away from Figure 10 and 12 is that the adjustments in PER at the time of the earnings release are not necessarily made toward the historical trend. The downward adjustments in 2003, 2005, and 2010 represent the cases where the aboveexpectation earnings were released or optimistic speculations with momentum-cause were in place, before the announcement of the earnings. The upward adjustments in 2004 and 2009 represent the cases where the below-expectation earnings were released or pessimistic speculations with momentum-cause were in place (De Long, Shleifer, Summers, and Waldmann, 1990). The adjustments at the point of earnings releases do not necessarily minimize the deviation of PER from the historically trending PER. This 11 20110429 Minseok Lee phenomenon is likely to be attributable to those who execute trades after reaffirming their own convictions on the traded shares. Figure 17. PER of KOSPI Based on The Annually Reported Earnings (Jan. 2001 – Dec. 2010, Monthly) 50 45 200103 40 200204 35 200909 30 200905 25 20 15 10 5 201003 200503 200107 200304 200305 200505 200710 200704 200903 201005 200101 200105 200109 200201 200205 200209 200301 200305 200309 200401 200405 200409 200501 200505 200509 200601 200605 200609 200701 200705 200709 200801 200805 200809 200901 200905 200909 201001 201005 201009 0 Source: Bloomberg LP *Reminder: Earnings’ releases occur in Aprils. In the foreign exchange market, any large amount of currency flows with profitcause is described as hot money, as there is no clear measure for the fair values of currencies (Razin and Rosefielde, 2011). Short-term speculative capital flows from abroad, or hot money, are believed to increase volatility, as seen in China (Guo and Huang, 2010). Being one of the less developed markets with relatively small capital pool and high domestic economic growth, the KOSPI and won-dollar exchange markets are not exceptions in terms of being affected significantly by foreign investment flows. For Korea’s case, insufficient amount of analysis and discussion has been made on whether or not the portfolio flow of foreign origin in aggregate is hot money. 12 20110429 Minseok Lee III. Analysis of The Residual PER (RPER)’s Impact on Trading Flow In the KOSPI, there is a popular stereotyping of foreign investors that they are predatory and driven by momentum. This analysis observes how true this perception has been in practice. More specifically, the author analyzes the behavior of each investor group to find out whose value-orientation is the strongest by measuring how much each investor group rely their trades on the gap between the historically trending PER and the PER of the time, with an anticipation that the deviation will disappear. This analysis measures the impact of the current PER’s deviation from the historically prevailing PER on each investor group’s trading behaviors. Those who add momentum toward the historically prevailing PER are seen as more value oriented investors. III. 1. Design III. 1. (1) Big Picture Various approaches at different dimensions are possible in the analysis of the traded equity markets. While the author could have used the variables at different facet, such as: interest rates, unemployment rates, and base commodity prices, he chooses to focus on a dimension that simplifies the analysis significantly. The author designs the experiment with a reasonable assumption that trade executions are necessarily made when the investors want, willingness, and are capable of, constraint, making the trades (Figure 13). 13 20110429 Minseok Lee Figure 13. Conditions For Trade Executions Willingness includes (1) a controllable willingness (valuation) and (2) an uncontrollable willingness (risk appetite).10 The above distinction between controllable and uncontrollable willingness must be made in this analysis, as it is clear that two different sorts of willingness exists. Valuation-levels are identified after investors’ research and analysis, and the investors are likely to choose whether or not to purchase or sell shares with logic; however, risk appetites carry emotional component in it, and therefore it is not modulated by the investing selves (Lee, Shleifer and Thaler, 1990). For the measure of the risk appetites, Chicago Board Options Exchange (CBOE) Market Volatility Index (VIX) is included.11 Constraints include (1) availability of funds (market liquidity) and (2) regulatory factors. To analyze foreign investors’ situation, the availability of funds is taken into account with two variables: won-dollar exchange rate (KRWD) and USD London Interbank Offered Rate (LIBOR) spreads between two different maturities. Won-dollar exchange rate is paired with USD LIBOR yield spreads between 1 months’ and 3 months’ (SPREAD). The USD LIBOR yield spreads represents anticipated short-term 10 Both valuation and risk appetite are motivations for trading; however, unlike valuation that serves as an opportunity that the investors intend to exploit, risk appetite is not managed by the investors’ will. 11 VIX “is a good indicator of the level of fear or greed in US and Global capital markets. When investors are fearful, the VIX level is significantly higher than normal” (Traub, Ferreira, McArdle, and Antognelli, 2000). Two additional reasons support the use of VIX in this context: (1) Korea’s economy’s structural vulnerability to overseas markets due to its focus on exports, and (2) the high correlation among major equity indices around the world (Figure 9). 14 20110429 Minseok Lee liquidity in the market. USD LIBOR 1-month yield and 3-month yield are commonly believed to be sensitive to almost the identical variables, except that the uncertainties for the two-month-long gap makes the 3 months’ rate slightly higher in the normal situations. The won-dollar exchange rate displays the currency impacts, as when the won weakens vis-à-vis the USD South Korean equities seem discounted in the foreign investors’ perspective, and vice versa. No significant enough regulatory factors exist, as elaborated in Background 3; therefore, the impact of regulatory factors is zero. While the point made by numerous researchers that risk appetite and macroeconomic uncertainties are two major variables that affect asset pricing, are well taken,12 macroeconomic uncertainties whose impact is directly reflected on the reported earnings was not included in this model as the PER is calculated from the reported earnings. The author adds the KOSPI’s total market turnover, as an additional control variable. Total market turnover is the sum of the products of each share’s price and the volume, and the unit is Korean Won. The inclusion of the turnover enables the author to better capture the trades’ impact on the KOSPI’s fluctuation than otherwise, as the traded amount can differ significantly in different periods. Figure 14. Summary of Variables Used in Panel A N = = Net Purchase Monetary Amount of Each Investor Group in KRW ƒ (RPER, VIX*, SPREAD, KRWD, Turnover) RPER = SPER** – IPER*** SPREAD = USD 3M LIBOR Yield – USD 1M LIBOR Yield KRWD = Exchangeable Amount of KRW per 1 USD *Detailed description of the VIX can be found in CBOE’s white paper (CBOE, 2011). **SPER: adjusted and smoothed PER. calculation of SPER is described in detail in III. 1. (2) Detailed Scene. ***IPER: Index PER. The calculation of IPER is described in detail in III. 1. (2) Detailed Scene. 12 Bansal and Yaron (2004), and Bekaert, Engatrom, and Xing (2009) 15 20110429 Minseok Lee III. 1. (2) Detailed Scene First, the author identifies the net purchased-sold monetary amount in Korean Won of each investor group (NET) as a dependent variable.13 Then, the residual PER (RPER) is calculated by calculating and deducting the index PER (IPER) from the adjusted/smoothed PER (SPER) (Figure 100, 11 and 12).14 The residual PER is used as the independent variable that represents the adjusted/smoothed PER (SPER)’s deviation from the historically prevailing SPER (IPER). The adjustments and smoothing of PER is a necessary step, as the use of SPER enables the analysis to capture the effects of prices and earnings, evenly. This adjustment is necessary because companies’ annual earnings are officially announced once a year, while the price fluctuates much more frequently. In the KOSPI, the time of earnings announcements are concentrated around April, and rapid adjustments in PER occur in Aprils, as described in Background 3 and in Figure 16 and 17. For each investing entity, value-orientation is based on their knowledge on the earning’s capability of the target company for the investment and the share valuation is supposed to be at the reasonable level at the individual level. Under such circumstances, because these investors’ view and knowledge can change dramatically over time, the use of SPER is once more justified. Due to the lack of market wide earnings’ data at a higher frequency, estimation of monthly earnings based on the annual change and the smoothing of earnings were conducted. 13 The resulting terms are NETFOREI (Nf), NETINSTI (Ni), NETINDIV, and NETOTHER (No) (Figure 15). Among various valuation metrics, the PER, one of the two (the other being the price to book ratio) most widely accepted standard of valuation in contemporary financial industry in Korea, was chosen. Also, the fact that PER is the only financial ratio (valuation metric) audited by accounting firms makes it more reliable. The need for the adjustments and smoothing on the PER is explained in III. 1. (2) Detailed Scene. 14 16 20110429 Minseok Lee ‘Estimation of monthly earnings based on the annual change’ of reported earnings looks at the path of earnings based on the earnings in Mays. As shown in figure 16, the updated earnings’ impacts are in place from May to the April of the next year, as the perceived earnings’ levels are set in Mays, as the earnings tend to level out from May to April. Also, it is logical to assume that the earnings numbers obtained in April is gradually outdated. Therefore, the earnings’ number of May is most reliable. Adjusted earnings (A_E) are the resulting values of this estimation (Figure 144). In the estimation process, the author makes an assumption that the amount of earnings of the next year (May) becomes clearer as the next May approaches, and gave weights according to this logic (Figure 100). The weights are given as in Figure 100 because there is a high likelihood that investors will have better understanding of the earnings’ situation based on either information leakage or research, as the time of earnings announcement approaches. After identifying the adjusted earnings (A_E), the author generates smoothed earnings (S_E) that is more useful for the anlysis than A_E is. In the smoothing process with average, the author included 2 months prior to the month and the following 10 months. Such design is made to reflect the fact that equity investors in general are forward-looking and there are a lot more focus on the predictions than on the past record, in their research. Inclusion of 2 months prior to the month is made because the past earnings can affect the investors’ confidence in the short term. This attributes to the fact that the history of companies could be a barometer of the quality of the business for some investors. 17 20110429 Minseok Lee Smoothed PER (SPER) is calculated using the monthly market capitalization of the KOSPI and the smoothed earnings (S_E), and Index PER is calculated as shown in Figure 11. Figure 100. Weights Used in The Estimation of Adjusted Earnings (A_E) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 5 6 7 8 9 Weight on the May of Y 10 11 12 1 Weight on the May of Y+1 2 3 4 Month 5 Figure 16. Perceived Earnings* of KOSPI (Jan. 2001 – Dec. 2010, Monthly) 8.00E+07 KRW 7.00E+07 6.00E+07 5.00E+07 4.00E+07 200109 200505 200705 200107 200305 1.00E+07 200103 201005 200804 201003 200503 200101 200105 200109 200201 200205 200209 200301 200305 200309 200401 200405 200409 200501 200505 200509 200601 200605 200609 200701 200705 200709 200801 200805 200809 200901 200905 200909 201001 201005 201009 0.00E+00 200405 200303 200204 200208 200904 200905 200403 3.00E+07 2.00E+07 200704 200805 Source: Bloomberg LP *KOSPI’s Perceived Earnings = Total Market Capitalization / KSPEKOSP Index **KSPEKOSP Index measures the PER based on the reported earnings. However, because the earnings were estimated with help of total market capitalization in the equation above, KOSPI’s earnings perceived by investors is calculated. 18 20110429 Minseok Lee Figure 144. Adjusted Earnings and Smoothed Earnings Based on Aprils’ Reports (May 2003 – May 2010, Monthly) 7.00E+13 6.00E+13 5.00E+13 4.00E+13 3.00E+13 2.00E+13 1.00E+13 200305 200309 200401 200405 200409 200501 200505 200509 200601 200605 200609 200701 200705 200709 200801 200805 200809 200901 200905 200909 201001 201005 0.00E+00 A_E S_E *A_E: Adjusted Earnings, S_E: Smoothed Earnings Figure 11. Calculation of Index PER (IPER) and Residual PER (RPER) (Jul. 2003 – Aug. 2009, Monthly) Given that: Time Index (TINDEX): 1 to 74, respectively, from Jul. 2003 to Aug. 2009, Solving for: SPER = Slope × TINDEX + SPER-intercept, Slope SPER-intercept Coefficients 0.0824472 8.861607 t 8.07 20.11 Index PER (IPER) = 0.0824472× TINDEX + 8.861607 Residual PER (RPER) = SPER - IPER 19 P > abs (t) 0.000 0.000 20110429 Minseok Lee Figure 12. Smoothed PER (SPER) and Index PER (IPER) (Jul. 2003 – Aug. 2009, Monthly) 18 16 14 12 SPER IPER 10 8 200307 200311 200403 200407 200411 200503 200507 200511 200603 200607 200611 200703 200707 200711 200803 200807 200811 200903 200907 6 Source: Bloomberg LP *SPER = Monthly KOSPI Total Market Capitalization / S_E **Calculation of IPER can be found in Figure 11. Figure 1000. Residual PER (RPER) 5 4 3 2 1 -1 -2 200307 200310 200401 200404 200407 200410 200501 200504 200507 200510 200601 200604 200607 200610 200701 200704 200707 200710 200801 200804 200807 200810 200901 200904 200907 0 -3 -4 Source: Bloomberg LP *Reminder: RPER = SPER - IPER 20 20110429 Minseok Lee Figure 15. Panel A Structure Time 0 . . . t 0 . . . t 0 . . . t 0 . . . t Dependent Variable Independent Variable Control Variables NFOREI (Nf) RPER VIX SPREAD KRWD Turnover NINSTI (Ni) RPER VIX SPREAD KRWD Turnover NINDIV RPER VIX SPREAD KRWD Turnover NOTHER (No) RPER VIX SPREAD KRWD Turnover The author recognizes that there is a high probability of individual investors’ positive or negative sentiments having an impact on the local mutual funds due to the fact that mutual funds are commonly known to adjust their positions based on the fund flows that are initiated by individuals who hold accounts in the mutual fund. This possibility is supported by the correlation (-0.4165) between the daily net purchase-sold amounts of individuals and mutual funds, in the period from 2001 January to 2010 December. Strongly negative correlation between the mutual funds’ and individuals’ net purchasesold amount can be explained by the fact that there are two accessible ways for individuals to invest in equities: either buying mutual funds (recorded as mutual funds) or trading with personal trading accounts (recorded as individuals). Negative correlation implies that the author’s original conjecture, that mutual funds’ trading behavior might be affected by the individual investors’ sentiments (positive or negative), is likely to be invalid. Monthly net purchased-sold amounts of the two show significantly low correlations (0.0614). The drastic change in correlations between the mutual funds and 21 20110429 Minseok Lee the individuals from daily and monthly data sets is likely attributable to the fact that the mutual funds increase or decrease their cash holdings to minimize the possible shock from the retail investors’ purchases and sales of the mutual fund account. As the correlations: -0.4165 and 0.0614 15 indicate, the longer time the mutual funds have, the better prepared they are for the shock from the individual investors who add or reduce their mutual fund accounts. The regressions in this paper use monthly data; therefore, the author decides that including the mutual funds under the institutional investors makes a better analysis. When determining the time duration of the analysis, the author chose to include only the period from Jul. 2002 to Dec. 2010, despite the availability of a complete panel data in the first half of 2002. Such decision was made in order to eliminate the noise, as the PER deviates excessively from the historical trend during the first half of 2002 (Figure 10 and 17). In addition, in the process of adjusting and smoothing of the earnings data, the observation period shrinks again from, Jul. 2002 to Dec. 2010, to, Jul. 2003 to Aug. 2009. The author recognizes that one of the most important motivations of value investment is the expectation for growth in future earnings. The reported-earnings-based PER is the data used for this analysis, and some readers may be concerned that the PER used in this paper does not capture the expectation for the earnings growth. However, the smoothing process used in the generation of SPER eliminated such concern. The author assumes that those investors who base their valuation analysis on the prediction of future earnings get the numbers correct with marginal errors. This assumption justifies the use of the reported-earnings-based PER as a valid valuation metric. 15 Source: KRX via Daewoo Securities Co. Ltd. 22 20110429 Minseok Lee III. 2. Results III. 2. (1) Preliminary Results The preliminary regression for foreign investors’ investment flow is made in Figure 18. The result of the preliminary regression shows that investment flows of foreign origin (KSFINET)16 is sensitive to the VIX, KRWD and Turnover. The reason why the impact from RPER is not clearly observed is likely to be attributable to the fact that the investors within each group are not monolithic. SPREAD’s unclear relevance to the net foreign portfolio flow is due to the variety of investor characters in terms of location and investment goals (Figure 19 and 20). What we see in the result section is the aggregate outcome. Also, the result of the preliminary regression represents the situation in both KOSPI and KOSDAQ, unlike the actual regression (only KOSPI). Figure 18. Estimation Results for KOSPI & KOSDAQ (Jul. 2003 – Aug. 2009, Collapsed to Monthly from Daily, Robust Regression) Parameters and the Estimates Time Series A: Variable KSFINET = π0 +π1RPER +π2VIX +π3SPREAD +π4KRWD + π5Turnover N/A π0 -2.00e+10 (-6.73) RPER π1 -9.76e+07 (-0.68) VIX π2 -2.14e+08 (-5.71) SPREAD π3 -1.57e+09 (-0.75) KRWD π4 2.18e+07 (7.38) Turnover π5 5.63e-06 (0.63) Source: Bloomberg LP *KSFINET’s unit is Million USD. **The values in the parenthesis indicate the t-statistics. ***Statistically significant Figure 90. Means and Elasticities From The Preliminary Run (Jul. 2003 – Aug. 2009, Collapsed to Monthly from Daily, Robust Regression) KSFINET RPER VIX SPREAD KRWD Turnover Mean -6.08E+08 1.78E-06 20.1324 0.170958 1075.092 8.27E+13 Elasticity N/A 2.86E-07 7.09E+00 4.41E-01 -3.85E+01 -7.66E-01 Figure 19. Composition of Foreign Investors by Number of Accounts 16 KSFINET Index: Daily Net Flows in South Korean equities by foreign investors in KOSPI and KOSDAQ. 23 20110429 Minseok Lee Category Individuals Institutions (Group Investments) (Pension Funds) (Brokerages) (Banks) (Insurers) (Unclear) End of 2008 End of 2009 End of 2010 29.29% 70.71% 41.44% 6.32% 2.37% 2.25% 1.34% 17.00% 29.01% 70.99% 41.27% 6.04% 2.30% 2.11% 1.28% 17.99% 28.76% 71.24% 41.85% 5.78% 2.21% 1.96% 1.24% 18.21% Source: Financial Supervisory Service Figure 20. Composition of Foreign Investors by Location (Jan. 2011) Location US Saudi Arabia China Cayman Is. Ireland Switzerland Luxemburg Hong Kong Denmark UAE Sweden Kuwait Germany Malaysia Virgin Is. Australia Canada Singapore Spain Japan France Netherlands UK Etc Purchased 71005 8916 4482 21756 11508 8349 19385 3800 1194 1132 1561 1424 9644 1860 1215 2917 3184 7593 396 2180 8444 4720 62514 6963 Sold 43869 5420 1528 19839 9651 6934 18577 3095 967 1023 1506 1526 9777 2136 1692 3467 3736 8814 1646 4218 13594 11456 74505 6866 Net 27136 3496 2954 1917 1857 1415 808 705 227 109 55 -102 -133 -276 -477 -550 -552 -1221 -1250 -2038 -5150 -6736 -11991 97 Source: Financial Supervisory Service *Unit for Purchased, Sold, and Net: 100 Million KRW. **Transaction’s Weight = (Purchased + Sold) / (Sum of all of Purchased and Sold) III. 2. (2) Thorough Analysis 24 Transaction's Weight 22.0% 2.7% 1.2% 8.0% 4.1% 2.9% 7.3% 1.3% 0.4% 0.4% 0.6% 0.6% 3.7% 0.8% 0.6% 1.2% 1.3% 3.1% 0.4% 1.2% 4.2% 3.1% 26.2% 2.6% 20110429 Minseok Lee In Figure 21, the author extensively utilizes dummy variables and interaction terms. Row 1 enables us to identify which investor trade on the other side of one another. Although Row 2 did not show statistically significant results, the row was meant to enable us to gauge each investor group’s trade execution’s sensitivity to RPER. In addition, rows 5 and 6 look at the KRW’s and Turnover’s impact on the NET, respectively. Because data used for the control variables were the same across the investor groups, use of interaction terms were applied for the control variables, as well. The VIX’s interaction terms with FOREI, INSTI, and OTHER: VIXf, VIXi and VIXo, read the VIX’s influence on foreign and institutional investors and other corporation, respectively (Row 3). Although not discussed afterwards due to the result’s statistical insignificance, SPREAD’s interaction terms with FOREI, INSTI, and OTHER: SPREADf, SPREADi, and SPREADo, were meant to measure the 1 Month and 3 Month LIBORs’ yield spread’s impact on each investor group’s trade executions (Row 4). To test for the presence of autocorrelation, Durbin-Watson d-statistics was used. As the resulting value (1.987788) is close to 2, we conclude that this regression has no autocorrelation issue (Figure 21). Figure 21. Estimation Results for KOSPI (Jul. 2002 – Dec. 2010, Monthly, Robust Regression) Parameters and the Estimates 25 20110429 Minseok Lee Panel A: N Variable = α1 + β1RPER + γ1VIX + δ1SPREAD + ε1KRWD + ζ1Turnover N/A α1 2.50e+12 (0.96) Interactions With FOREI Interactions With INSTI Interactions With OTHER αf -2.31e+13 (-5.59) αi 1.14e+13 (2.82) αo 1.68e+12 (0.62) + αfFOREI + βfRPERf + γfVIXf + δfSPREADf + εfKRWDf + ζfTurnoverf RPER β1 8.39e+10 (0.81) βf -1.50e+11 (-0.83) βi -1.10e+11 (-0.73) βo -7.24e+10 (-0.67) VIX γ1 4.94e+10 (1.99) γf -2.69e+11 (-5.93) γi 8.97e+10 (2.26) γo -1.77e+10 (-0.64) + αiINSTI + βiRPERi + γiVIXi + δiSPREADi + εiKRWDi + ζiTurnoveri SPREAD δ1 1.38e+11 (0.08) δf -1.72e+12 (-0.62) δi 8.08e+11 (0.32) δo 3.91e+11 (0.22) + αoOTHER + βoRPERo + γoVIXo + δoSPREADo + εoKRWDo + ζoTurnovero KRWD ε1 -3.70e+09 (-1.52) εf 2.60e+10 (6.55) εi -1.08e+10 (-2.86) εo -3.62e+08 (-0.14) (Row 1) (Row 2) (Row 3) (Row 4) (Row 5) (Row 6) Turnover ζ1 0.0035631 (0.47) ζf 0.0063798 (0.57) ζi -0.0156617 (-1.29) ζo -0.0054136 (-0.69) Durbin-Watson d-statistics (NEW VALUE, NEW VALUE) = NEW VALUE *The values in the parenthesis indicate the t-statistics. **Statistically significant Figure 22. Means and Elasticities P RPER (Mean: 1.78e-06) VIX (Mean: 2.01e+01) SPREAD (Mean: 1.71e-01) KRWD (Mean: 1.08e+03) Turnover (Mean: 8.27e+13) Q Q’s Elasticity on P NFOREI NINSTI NINDIV NOTHER -1.27E-04 -5.01E-05 1.61E-04 2.21E-05 NFOREI NINSTI -4.77E+03 3.02E+03 NINDIV NOTHER 1.07E+03 6.88E+02 NFOREI NINSTI NINDIV NOTHER -2.92E+02 1.74E+02 2.54E+01 9.76E+01 NFOREI NINSTI NINDIV NOTHER 2.59E+04 -1.68E+04 -4.29E+03 -4.71E+03 NFOREI NINSTI 8.87E+02 -1.08E+03 3.18E+02 -1.65E+02 NINDIV NOTHER *Shaded parts indicate the results that are more reliable than the rest, due to the statistical significance. The darker, the more reliable. ** Calculations of the elasticity were done by the following equation: Elasticity = Coefficient × (Mean of RPER, VIX, SPREAD, KRWD, or Turnover) / (Mean of N). Mean of N is 9.27e+08. The result from the row 1 shows: α1 + αf (-2.06e+13) < 0 < α1 (2.50e+12) < α1 + αi (1.39e+13). The negative sign on α1 + αf indicates that the foreigners are on the other side 26 20110429 Minseok Lee of individual investors’ trade, while institutional investors and other corporations are on the same side of individual investors’ trade in aggregate. We intended to distinguish from this result which group has been trading in order to achieve the historically prevailing PER. The result shows no statistically significant results for the row 2. From the elasticity table (Figure 22), we identify that the foreign investors on average increased long positions (or decreased short positions) in the KOSPI by 684% when the VIX hiked 1%, while institutional investors increased short positions (or decreased long positions) in the KOSPI by 157%. Individual investors increased short positions (or decreased long positions) in the KOSPI by 6.65% when the VIX hiked 1%, ceteris paribus. In addition, the elasticity of KRWD’s NFOREI show that foreigners on average decreased long positions in the KOSPI by -3360% as KRW depreciated by 1%, ceteris paribus. KRWD’s NINSTI shows that 825% increase in long positions (or decrease short positions) were made by institutional investors for KRW’s every 1% depreciation vis-à-vis USD. As NINDIV indicates, individual investors increased long positions (or decreased short positions) by 24.6% for every 1% depreciation of KRW against USD. The result from the VIX might be misinterpreted to mean that foreign investors are risk-loving. More thorough observation on the trading flow and VIX can give more correct picture. In Figure 24, we can see that the VIX has been relatively high from 2007 to 2009 (and 2010).17 From the beginning of the subprime mortgage crisis of 2007 to the post-Lehman period of the end of 2009, individual investors and foreign investors show two distinct behaviors. Until the Oct. 2008, the VIX hiked rapidly (Figure 24), and we observe that the foreigners have aggressively reduced the KOSPI in 2007 and 2008 17 Figure 27 does not contain the data in 2010, so I exclude 2010 from the analysis. 27 20110429 Minseok Lee (Figure 27). Unlike the foreign investors, domestic individual investors were net buyers of the KOSPI in 2007 and 2008. In 2009, although the VIX stayed at a high level, it headed down at a fast pace (Figure 24). During this period, foreign investors added significant amount of the KOSPI in aggregate. This observation indicates that the direction of the VIX movement is a more important indicator than the level of the index, as stabilizing events encourage investors to make optimistic speculation.18 This finding, with help of knowledge acquired from the regression of the Panel A, indicates that foreign investors do not mind adding the KOSPI even as the VIX rises, as long as they expect the VIX to be lower in future. This means that foreign investors are more concerned about the risk managements than individual investors. The result of this analysis weakens the argument that foreign investors utilize depreciations of won as buying opportunities. What this analysis implies is that foreigners’ investment flows caused by factors other than currencies, such as liquidity squeeze, affects the currency market. LIBOR spread’s spike in the period between Nov. 2008 and Jul. 2009 (Figure 23), and its relation to the spike in the foreign exchange market is worth noting (Figure 25). This provides a good explanation with regard to the simultaneous local currency depreciation and the outflow of the portfolio flow of foreign origin: as foreigners sell Korea, they abandon Korean Won at the same time. Figure 17 shows that the PER level hiked during the same period, and this implies that foreign investors’ purchases of the KOSPI were caused by depressed Korean Won. The observations in Figure 25 and 28 show that under the circumstances where the local currency (KRW) is extremely depressed, foreign investors’ investment inflow The author classifies this as a speculation, as the KOSPI’s earnings perceived by investors was extremely poor in 2009 (Figure 16). 18 28 20110429 Minseok Lee occurs to Korea. The period from Sept. 2008 to Jul. 2009 can be classified as the time period where KRW was extremely depressed (Figure 25), and its impact on the portfolio flows of foreign origin is very clear during the same period (Figure 28). Figure 23. LIBOR Yield Spread between 1 Months’ and 3 Months’ (SPREAD) (Jul. 2002 – Dec. 2010, Averaged Monthly From Daily) 0.9 200901, 0.827411 0.8 0.7 0.6 0.5 201006, 0.1881623 0.4 0.3 0.2 0.1 -0.1 200207 200211 200303 200307 200311 200403 200407 200411 200503 200507 200511 200603 200607 200611 200703 200707 200711 200803 200807 200811 200903 200907 200911 201003 201007 201011 0 Source: Bloomberg LP *Calculation of SPREAD is explained in Figure 14. Figure 24. VIX (Jul. 2002 – Dec. 2010, Averaged Monthly From Daily) 29 20110429 Minseok Lee 70 200810 60 50 201005 40 30 20 10 201011 201007 201003 200911 200907 200903 200811 200807 200803 200711 200707 200703 200611 200607 200603 200511 200507 200503 200411 200407 200403 200311 200307 200303 200211 200207 0 Source: Bloomberg LP *For the VIX, the magnitude of swing is the major concern, as oppose to the numerical level (vertical axis) of the VIX. Figure 25. Won-dollar Exchange Rate (KRWD) (Apr. 2000 – Dec. 2010, Daily) 1800 10/24/2008, 1454.5 1600 2/27/2009, 1538.3 1400 1200 9/29/2008, 1184.9 1000 800 4/25/2007, 926.8 600 7/9/2009, 1279 400 200 0 Source: Bloomberg LP *KRWD = Exchangeable Amount of KRW per 1 USD Figure 26. Change in Shareholding of KOSPI by Investor Group (2001 – 2009, Annually) 30 5/25/2010, 1271.68 20110429 Minseok Lee Year 2001 2002 2003 2004 2005 2006 2007 2008 (%) 2009 6.43 -6.29 2.28 -0.61 -2.38 0.01 4.1 -0.27 -2.63 1.86 0.8 -1.7 -2.24 1.54 0.43 -2.47 2.67 -0.49 -4.89 -1.95 3.85 -3.59 -9.03 5.26 3.87 -0.9 3.97 OTH -2.44 Source: KRX 2.99 -1.2 -0.95 0.27 0.28 2.98 7.36 -6.93 Investor Group FOR INS IND Figure 27. Change in Shareholding of KOSPI by Investor Group (2001 – 2009, Annually) 10 % 8 6 4 2 0 2001 2002 2003 2004 2005 2006 2007 2008 -2 -4 -6 -8 -10 FOR INS IND Source: KRX Figure 28. KSFINET Index USD (Jan. 2001 – Dec. 2010, Monthly) 31 OTH 2009 20110429 Minseok Lee 6E+09 200704, 2781500000 4E+09 201004, 4866220000 200805, 878080000 2E+09 -2E+09 200101 200105 200109 200201 200205 200209 200301 200305 200309 200401 200405 200409 200501 200505 200509 200601 200605 200609 200701 200705 200709 200801 200805 200809 200901 200905 200909 201001 201005 201009 0 -4E+09 -6E+09 -8E+09 -1E+10 -1.2E+10 200708, -9597800000 201005, -5423890000 200801, -9474960000 Source: Bloomberg LP *KSFINET Index: Daily Net Flows in South Korean equities by foreign investors in KOSPI and KOSDAQ. Figure 26. KOSPI Index (Jul. 2002 – Dec. 2010, Daily) 2500 2000 1500 1000 500 200207 200211 200303 200307 200311 200403 200407 200411 200503 200507 200511 200603 200607 200611 200703 200707 200711 200803 200807 200811 200903 200907 200911 201003 201007 201011 0 Source: Bloomberg LP Above interpretations of results are valid within the period from Jul. 2002 to Dec. 2010; however, the result from the following analysis indicates that each investor group’s 32 20110429 Minseok Lee trading behavior changes over time. The author examined the varying investment approach of each investor group over time by running the regression of Panel A (Figure 21) with additional dummies for two periods: the first half (FH=1) and the second half (FH=0). In Figure 30, FOREIfh, INSTIfh, OTHERfh, RPERfh, RPERfh,f, RPERfh,i, RPERfh,o, VIXfh, VIXfh,f, VIXfh,i, VIXfh,o, SPREADfh, SPREADfh,f, SPREADfh,i, SPREADfh,o, KRWDfh, KRWDfh,f, KRWDfh,i, KRWDfh,o, Turnoverfh, Turnoverfh,f, Turnoverfh,i, and Turnoverfh,o indicate the interaction terms between FH and FOREI, INSTI, OTHER, RPER, RPERf, RPERi, RPERo, VIX, VIXf, VIXi, VIXo, SPREAD, SPREADf, SPREADi, SPREADo, KRWD, KRWDf, KRWDi, KRWDo, Turnover, Turnoverf, Turnoveri, and Turnovero, respectively. The F-test’s result in Figure 30 shows that each investor’s characteristics that are identified from Figure 21 are applicable only for the period from Jul. 2002 to Dec. 2010. 19 This is likely to be attributable to two major reasons: the changes in the composition of each investor group and each investing selves’ behaviors. The author was concerned that the outliers’ impacts were not interpreted in detail; hence, more tests followed (Figure 31). Both KRWD and VIX,20 appear to change their degree of impact on the trading behavior of foreign investors, as the hypothesis 2 and 4 were rejected in Figure 31. The elasticities of the two variables in the first and the second half also differ significantly (Figure 32). Figure 30. Test of Change in Character within Each Investor Group Over Time (Jul. 2002 – Dec. 2010, Monthly, Robust Regression) Parameters and Estimators 19 This is in line with the findings from Cheo, Kho, and Stulz (1998) and Kim and Wei (1999). Two variables with statistically significant results from Figure 21, among those that affected the result of the F-test in Figure 30, are the VIX and KRWD. 20 33 20110429 Panel A: Variable Minseok Lee N = _α1 + _β1RPER + _γ1VIX + _δ1SPREAD + _ε1KRWD + _ζ1Turnover + ω00FH + ω10RPERfh + ω20VIXfh + ω30SPREADfh + ω40KRWDfh + ω50Turnoverfh N/A _α1 Interactions With FOREI Interactions With INSTI Interactions With OTHER Interactions With FH Interactions With FH, F Interactions With FH, I Interactions With FH, O 2.07e+12 (0.40) _αf -2.74e+13 (-4.25) _αi 1.64e+13 (2.43) _αo 2.88e+12 (0.55) + _αfFOREI + _βfRPERf + _γfVIXf + _δfSPREADf + _εfKRWDf + _ζfTurnoverf + ω01FOREIfh + ω11RPERfh,f + ω21VIXfh,f + ω31SPREADfh,f + ω41KRWDfh,f + ω51Turnoverfh,f RPER _β1 2.76e+10 (0.13) _βf -4.65e+11 (-0.92) _βi 3.80e+11 (1.00) _βo -3.93e+10 (-0.17) + _αiINSTI + _βiRPERi + _γiVIXi + _δiSPREADi + _εiKRWDi + _ζiTurnoveri + ω02INSTIfh + ω12RPERfh,i + ω22VIXfh,i + ω32SPREADfh,i + ω42KRWDfh,i + ω52Turnoverfh,i VIX _γ1 4.13e+10 (1.00) _γf -3.48e+11 (-5.42) _γi 1.90e+11 (3.15) _γo -6.23e+09 (-0.14) SPREAD _δ1 -8.74e+10 (-0.03) _δf -3.58e+12 (-1.05) _δi 2.90e+12 (0.86) _δo 1.14e+12 (0.43) + _αoOTHER (Row 1) + _βoRPERo (Row 2) + _γoVIXo (Row 3) +_ δoSPREADo (Row 4) + _εoKRWDo (Row 5) + _ζoTurnovero (Row 6) + ω03OTHERfh (Row 7) + ω13RPERfh,o (Row 8) + ω23VIXfh,o (Row 9) + ω33SPREADfh,o (Row 10) + ω43KRWDfh,o (Row 11) + ω53Turnoverfh,o (Row 12) KRWD _ε1 -3.54e+09 (-0.59) _εf 3.29e+10 (4.32) _εi -1.71e+10 (-2.22) _εo -1.87e+09 (-0.31) Turnover _ζ1 0.0086209 (1.08) _ζf 0.0038968 (0.20) _ζi -0.0314433 (-2.01) _ζo -0.0061355 (-0.71) ω00 ω10 ω20 ω30 ω40 ω50 -2.27e+12 (-0.38) 1.35e+11 (0.55) 2.31e+11 (2.33) 8.54e+12 (2.12) -1.68e+09 (-0.26) -0.0047034 (-0.39) ω01 ω11 ω21 ω31 ω41 ω51 2.32e+13 (2.67) 5.99e+11 (1.07) -2.85e+11 (-1.53) -1.22e+13 (-1.82) -1.64e+10 (-1.83) -0.025298 (-0.89) ω02 ω12 ω22 ω32 ω42 ω52 -1.42e+13 (-1.81) -9.33e+11 (-2.22) -4.75e+11 (-3.30) -1.34e+13 (-2.36) 1.98e+10 (2.40) 0.044402 (2.06) ω03 ω13 ω23 ω33 ω43 ω53 -2.17e+10 (-0.00) -1.87e+11 (-0.70) -1.73e+11 (-1.62) -8.66e+12 (-2.05) 3.64e+09 (0.55) -0.0011949 (-0.09) Hypothesis H0: ω00 ~ ω53=0 Result F (24, 248) = 2.93 Prob > F = 0.0000 Figure 31. Examination of Change in Character within Each Investor Group Over Time Hypotheses F-Test Results 34 20110429 Minseok Lee (0) H0: ω00=ω01=ω02= ω03=0 F (4, 248) = 5.80 Prob > F = 0.0002 (Variable: N/A) (1) H0: ω10=ω11=ω12= ω13=0 F (4, 248) = 0.02 Prob > F = 0.0915 (Variable: RPER) (2) H0: ω20=ω21=ω22= ω23=0 F (4, 248) = 3.29 Prob > F = 0.0119 (Variable: VIX) (3) H0: ω30=ω31=ω32= ω33=0 F (4, 248) = 1.61 Prob > F = 0.1719 (Variable: SPREAD) (4) H0: ω40=ω41=ω42= ω43=0 F (4, 248) = 5.49 Prob > F = 0.0003 (Variable: KRWD) (5) H0: ω50=ω51=ω52= ω53=0 F (4, 248) = 1.98 Prob > F = 0.0985 (Variable: Turnover) Figure 300. Variables’ Means MEAN N RPER VIX SPREAD KRWD Turnover 9.27E+08 1.78E-06 20.1324 0.1709575 1075.092 8.27E+13 Figure 32. Comparison of the Elasticity of The First and The Second Half of The Period IV or CV** FOREI RPER VIX SPREAD KRWD Turnover -8.40E-04 -6.66E+03 -6.76E+02 3.41E+04 1.12E+03 First Half of The Period INSTI INDIV 7.83E-04 5.02E+03 5.19E+02 -2.39E+04 -2.04E+03 -8.93E-04 8.97E+02 -1.61E+01 -4.11E+03 7.69E+02 OTHER FOREI -2.25E-05 7.62E+02 1.94E+02 -6.27E+03 2.22E+02 1.41E-03 -1.17E+03 -6.75E+02 -2.10E+04 -2.68E+03 Second Half of The Period INSTI INDIV -1.53E-03 -5.30E+03 -8.96E+02 2.10E+04 3.54E+03 2.59E-04 5.02E+03 1.57E+03 -1.95E+03 -4.20E+02 OTHER -9.98E-05 1.26E+03 -2.21E+01 2.27E+03 -5.26E+02 **IV: Independent Variable, CV: Control Variable * indicates reliable results whose coefficients are statistically significant. IV. Conclusion Listed and traded equities are important segments of contemporary financial markets. Yet, it is unclear which factor validates the price level of the listed shares. In the absence of unclear universal guideline in equity markets, the answer for the following 35 20110429 Minseok Lee question: which factors determine most sophisticated investors’ (foreign investors) portfolio flow, is sought in this paper. This paper does not identify clear evidences that value-orientation is a major concern for all investors, and concludes that it is unclear whether foreign investors’ trade executions are momentum or value-caused. However, in the process, we have positively identified that foreign investors react sensitively to the VIX. The risk-averse characteristic of foreign investors identified in this paper shows that claims that foreign investors manipulate the market with their funding capability are falsifiable. The variety in sources of funds in terms of character and location also makes the manipulation argument weak. From this aspect, foreign investors take a constructive role for the domestic equity market’s development, as their risk-averse attitude and intelligent estimates on the future risk add value to the KOSPI. Additionally, this paper concludes that foreign investors’ sales-purchases of Korean equities and the Korean Won are simultaneously done, and when the Korean Won depreciates excessively, foreign investors’ trade executions are affected by their anticipation on the Korean Won that the historical norm of the Korean Won is maintained. This is stability that foreign investors add to the won-dollar market, albeit contributing to the elevated KOSPI’s valuation in the short term. The conclusions above are applicable to the specified window of Jul. 2002 to Dec. 2010. While the preferences and composition of investors shift over time, the analyses demonstrated in this paper provides an accurate snapshot of the motivations of each investor group from the substantive period of Jul. 2002 to Dec. 2010. Thus, this paper 36 20110429 Minseok Lee frames the assumptions that drive decisions made by foreign investors in the KOSPI to facilitate a clearer understanding of the developments to date. References Bansal, R. and A. Yaron (2004), “Risks for the Long Run: A Potential Resolution of Asset Pricing Puzzles,” Journal of Finance, Vol. 59, pp. 1481-1509. Barber B. M., and T. Odean (2008), “All that glitters: The effect of attention and news on the buying behavior of individual and institutional investors.” Review of Financial Studies, Vol. 21, pp. 785-818. Bekaert, G., E. Engstrom, and Y. Xing (2009), “Risk, Uncertainty, and Asset Prices,” Journal of Financial Economics, Vol 91, pp. 59-82. Bohl, Martin T., Christian A. Salm, and Michael Schuppli (2011), “Price Discovery And Investor Structure In Stock Index Futures.” The Journal of Futures Markets, Vol. 31, No. 3, pp. 282-306. Choe, Hyuk, Bong-Chan Kho, and rene M. Stulz (1998), “Do Foreign Investors Destabilize Stock Markets? The Korean Experience in 1997.” NBER Working Paper, No. 6661. Chicago Board Options Exchange (CBOE), Incorporated (2011), “The CBOE Volatility Index®.” cboe.com. http://www.cboe.com/micro/vix/vixwhite.pdf (accessed April 11, 2011) Financial Supervisory Service (2011), “Foreign investors’ trading trend in Korea in Jan. 2011.” www.fss.or.kr. http://www.fss.or.kr/fss/kr/bbs/view.jsp?url=/fss/kr/120739703060 5&bbsid=1207397030605&idx=1298268925786&num=161 (accessed April 23, 2011) (in Korean) Frazzini, A. and O.A. Lamont (2008), “Dumb money: Mutual fund flows and the crosssection of stock returns.” Journal of Financial Economics, Vol. 88, pp. 299-322. Graham, Benjamin, and David L. Dodd (2008), The Security Analysis. 6th ed. USA: The McGraw-Hill Companies. Gompers, P.A. and A. Metrick (2001), “Institutional investors and equity prices.” Quarterly Journal of Economics, Vol. 116, pp. 229-259. Grinblatt, M. and M. Keloharju (2000), “The investment behavior and performance of various investor types: A study of Finland’s unique data set.” Journal of Financial Economics, Vol. 55, pp. 43-67. Guo, Feng, and Ying Huang (2010), “Hot Money and Business Cycle Volatility: Evidence from China.” China and World Economy, Vol 18, No. 6, pp. 73-89. 37 20110429 Minseok Lee Kim, Woochan, and Shang-Jin Wei (1999). “Foreign Portfolio Investors Before And During A Crisis.” NBER Working Paper Series, No. 6968. http://www.nber.org/papers/w6968.pdf Klarman, Seth A (2008), “The Timeless Wisdom of Graham and Dodd.” In The Security Analysis. 6th ed. The USA: The McGraw-Hill Companies. Korea Exchange (2011). Interview with the department representative. Phone interview. March 3, 2011. Lee, Charles M.C., Andrei Shleifer, and Richard Thaler (1990), “Anomalies. Closed-End Mutual Funds.” Journal of Economic Perspectives, Vol. 4, No. 4, pp. 153-164. Lee, Seyoung, and Alison Tudor (2011), "Seoul Punishes Deutsche for Market Manipulation." The Wall Street Journal, Feb 24, 2011. http://online.wsj.com/article/SB100014 24052748703775704576161870367843839.html (accessed April 7, 2011). De Long, J. Bradford, Andrei Shleifer, Lawrence H. Summers, and Robert J. Waldmann (1990), “Positive Feedback Investment Strategies and Destabilizing Rational Speculation.” Journal of Finance, Vol. 45, No. 2, pp. 379-395. Razin, Assaf, and Steven Rosefielde (2011), “Currency and Financial Criese of the 1990s and 2000s.” NBER Working Papers, No. 16754. Sias, R (2004), “Institutional Herding.” Review of Financial Studies, Vol. 17, pp. 165206. Traub, Heydon D., Luis Ferreira, Maria McArdle, and Mauro Antognelli (2000). “Fear and Greed in Global Asset Allocation.” The Journal of Investing, Vol. 9, No. 1, pp. 27-31. 38