The sources for our data are the databases (DB) maintained by the Taiwan Economic Journal (TEJ). They are the TEJ Equity DB, TEJ Finance DB, TEJ Company DB, and TEJ Macro DB. Our sample period is from the beginning of the third quarter of 1994 to the end of 2001. We use the following sample selection criteria to restrict our sample:
1. Firms must be listed for more than six months in order to give foreign institutional investors time to analyze and to make investment decisions;
2. Transaction and financial data must be available for the firms;
3. Foreign investment in the firms must be permitted. This filter excludes five onshore transportation firms (Stock Codes 2607, 2608, 2611, 2612, 2616) and one TV firm (Stock Code 9928).
The application of the filters produces a sample size of 523 firms for out analyses of foreign ownership effect. The sample size declines to 468 firms for our analyses of foreign equity ownership and firm characteristics. The smaller sample excludes 52 financial firms and three nonfinancial firms that lack adequate firm characteristic data. Table I, Panel A presents the number of firms in the sample by year.
In addition to the equity ownership data, we have data on non-QFII investor groups. Table I, Panel B presents the annual ownership percentages by investor type. On average, over 90% of the shares are owned by domestic individual and institutional investors. Foreign investors (QFII) account for a paltry 2.2% of market capitalization ownership on average, and there is little or no foreign ownership in half of the stocks. It may appear as though foreigner investment in Taiwan is unimportant. However, as we will discuss below, this interpretation is misleading. (15)
For Japan, Kang and Stulz (1997) demonstrate that the foreigner holdings are disproportionate across industries with a concentration in large, liquick export-oriented firms. Foreigners reveal a similar preference for certain industries in Taiwan. We document foreigner preferences for equal-weighted and value-weighted domestic firms using Gini coefficients. Figure 1 provides an illustration of plotting Lorenz curves for foreign ownership for equal-weighted domestic firms in 1995. A perfectly equal foreign ownership distribution is represented by the 45 degree straight line. The area between the 45 degree line and the convex Lorenz curve is a measure of the inequality or the concentration of foreigner investment. The Gini coefficient computes this area as a percentage of the total area under the 45 degree line. The coefficient increases from 62.9% in 1995 to 79.1% in 2001. We also compute the Gini coefficients for value-weighted domestic firms. They are lower than the equal-weighted firms and average about 47% during our sample period. The evidence is inconsistent with a passive investment strategy whereby foreigners allocate their funds to local stocks in the same proportion as that dictated by popular indices for the Taiwan market.
Panel C presents the descriptive statistics for our sample of nonfinancial firms' characteristics. All the firm characteristic variables and the characteristics used to construct the variables used in the paper are listed here. Daily return volatility is the standard deviation of daily returns and is calculated annually. For all other financial variables, we take the average of the annual data for each firm and then average them across all firms.
An issue related to firm ownership is the degree of ownership concentration. In markets with poor shareholder protection, firms are often controlled by a few large shareholders who are too powerful for other shareholders to monitor. Dahlquist, Pinkowitz, Stulz, and Williamson (2003) observe that ownership concentration is related to home bias. They demonstrate that the fraction of shares that can be traded freely helps to explain the extent of the home bias. In our analysis, we include a proxy for ownership concentration (inside ownership) that we measure as the fraction of shares owned by corporate insiders, defined as officers and members of the board of directors. This measure represents ownership by controlling shareholders. Panel C indicates that inside ownership in our sample averages 26.53% with a median of 24.84%. By way of comparison, Dahlquist et al. report that for the few firms for which Worldscope has closely held share data, Taiwan's ratio of world float portfolio to world market portfolio was 99.29% in 1997.
Panel D provides the correlation matrix of all the variables used in our analyses for nonfinancial firms. In general, the correlations are consistent with the results reported below. For example, the positive correlations between foreign ownership and measures of firm size (capitalization, total assets, and net sales) hint at foreign investors' preference for large firms. Similarly, the positive correlation between foreign ownership and the export ratio suggests foreign investors' preference for export-oriented firms. Additionally, the high correlation between Tobin's Q and both foreign ownership and return on assets, respectively, anticipate our subsequent inference that foreign ownership is associated with superior firm performance.
III. Foreign Ownership Effect
We begin with a description of the underlying empirical asset pricing model and portfolios used in our analyses in Section III.A. Section III.B presents the foreign ownership effect. Section III.C provides some robustness checks of the foreign ownership effect. Section III.D presents a discussion of the valuation result.
A. The Four-Factor Model and Foreign Ownership Portfolios
We adopt the Fama-French (1993, 1996) model and complement their three factors with a momentum factor. The works of Jegadeesh and Titman (1993) and Carhart (1997) suggest the importance of including the fourth factor. The four-factor model can be stated as
[R.sub.i,t] - [RF.sub.t] = [alpha] [[beta].sub.RMRF] [RMRF.sub.t] [[beta].sub.SMB] [SMB.sub.t] [[beta].sub.HML] [HML.sub.t] [[beta].sub.PR1YR] [PR1YR.sub.t] [[epsilon].sub.i,t], (1)
[FIGURE 1 OMITTED]
where [R.sub.i,t] is the return on portfolio i at month t, [RF.sub.t] is the risk-free rate, [RMRF.sub.t] is the return on excess market portfolio, [SMB.sub.t] is the return on factor-mimicking size portfolio, [HML.sub.t] is the return on factor-mimicking book-to-market equity portfolio, and [PR1YR.sub.t] is the return on Carhart's (1997) factor-mimicking portfolio for one-year return momentum. The constant term or is Jensen's alpha and [[beta].sub.RMRF], [[beta].sub.SMB], [[beta].sub.HML], and [[beta].sub.PR1YR] the are factor loadings of RMRF, SMB, HML, and PR1YR, respectively.
Estimation of Equation (1) requires a time-series regression of monthly foreign ownership portfolio excess returns on RMRF, SMB, HML, and PR1YR. We next describe the construction of the four factors followed by the construction of the foreign ownership portfolios. In constructing the four factors, only firms with ordinary common stocks that have been listed for at least two years on the Taiwan Stock Exchange (TSE) or the Taiwan OTC market are included in our portfolios. This excludes Taiwan Depositary Receipts, convertible bonds, units of beneficial interest, and newly listed securities.
The excess market portfolio return, RMRF, is computed as the monthly return on a value-weighted portfolio of all TSE and OTC stocks minus the one month time deposit rate offered by the Bank of Taiwan. (16) To obtain the size factor SMB and value factor HML, portfolios are formed on the basis of size and book-to-market. All TSE stocks are ranked by size as of the end of June of each year t from 1994 to 2001. Size, or market equity (ME), is calculated as share price multiplied by shares outstanding. The stocks are divided into two groups, small (S) and big (B), using the TSE median size to divide the observations. Book-to-market equity (BE/ME) is book common equity (BE) for the fiscal year ending in calendar year t - 1, divided by market equity (ME) at the end of December of year t - 1. Here, BE is the book value of stockholder's equity, plus balance sheet deferred taxes and investment tax credit, minus the book value of preferred stock. The fiscal year ends in December for most Taiwanese firms. The groups are formed by categorizing each of the two size ranked groups (S and B) into three book-to-market ranked groups: the bottom 30% (L for low), middle 40% (M for medium), and top 30% (H for high). This gives us a total of six size/book-to-market portfolios: 1) S/L, 2) S/M, 3) S/H, 4) B/L, 5) B/M, and 6) B/H. Finally, monthly value-weighted returns on the six portfolios are calculated from the beginning of July of year t to the end of June of year t 1, and the portfolios are rebalanced at the end of June of year t 1.
The size factor SMB and the value factor HML are computed for the six portfolios. The factor SMB is the difference between the simple average of monthly returns on the three small stock portfolios and on the matching big stock portfolios: 1) S/L - B/L, 2) S/M - B/M, and 3) S/H - B/H. The factor HML is the difference between the simple average of monthly returns on the two high BE/ME portfolios (S/H and B/H) and on the matching low BE/ME portfolios (S/L and B/L): 1) S/H - S/L and 2) B/H - B/L.
The fourth factor is the momentum factor, PR1YR. It is the difference between the equal-weighted average of firms with the highest 30% 11-month returns, lagged 1 month, minus the equal-weighted average of firms with the lowest 30% 11-month returns, lagged 1 month. The portfolios include all TSE and OTC stocks and are rebalanced monthly.
The foreign ownership portfolios are constructed as follows. At the end of each quarter from Q2 1994 to Q3 2001 (a total of 30 quarters), we sort all sample firms into five portfolios based on their foreign ownership percentage. Portfolio P1 consists of stocks with the highest foreign ownership and P5 consists of those with the lowest. We then calculate equal-weighted and value-weighted monthly returns for the five portfolios in the following three-month period before rebalancing the portfolios.
Table II, Panel A presents the summary statistics of the four factors needed to estimate Equation (1). Panel B presents the descriptive statistics of the firm characteristics for these five portfolios. They indicate that foreign institutions prefer large, export-oriented firms. These results are similar to those round by KS for Japan. Higher foreign ownership portfolios also exhibit higher accounting returns and Tobin's Q. Finally, Panel B shows that the combined P1 and P2 portfolios account for over 70% of the market capitalization.
Panel C presents the percentage foreign ownership summary statistics for the five foreign ownership portfolios over the eight-year period from 1994 to 2001. Whereas foreign institutional ownership appears to be low in the previous panels, Panel C demonstrates an increase over time with P1 presenting the largest increase. This trend suggests that tests of the foreign ownership hypotheses may be more apparent in recent years, during which time there has been greater foreign ownership. However, our inferences are based on the entire sample period.
Panel D presents the summary statistics of the five equal-weighted and value-weighted portfolio returns. They illustrate that the mean and median monthly returns decline with decreasing foreign ownership percentage. Indeed, the differences between P1 and P5 are startling, exceeding 100 basis points for equal-weighted returns and 195 basis points for value-weighted returns.
In addition to the Taiwan version of the Fama-French (1993, 1996) model, we experiment with the global and the US versions. However, the global and US versions produced very poor fits and were discarded in favor of the Taiwanese version. The poor results for the global version may reflect the lack of global market integration that is responsible for the home bias in the first place. Finally, although we present the results of the four-factor model, similar results are obtained using the Fama-French (1993, 1996) three-factor model. (17)
B. Foreign Ownership Effect
Table III presents the market impact results of foreign institutional ownership. The sample includes all 523 firms for which we have complete data. The five portfolios used in the table are based on their foreign ownership ranking, with P1 having the highest foreign ownership percentage. The table illustrates the monthly estimation results for equal- and value-weighted portfolios. It also reports the results of going long in P1 and short in P5 (P l-P5).
For both equal- and value-weighted portfolios, the high foreign ownership portfolio outperforms the low foreign ownership portfolio. Equal and value-weighted portfolios formed by going long in P1 and short in P5 (P1-P5) earn significant positive Jensen alphas. The magnitudes of the alphas are huge. For example, the value-weighted P1-P5 portfolio has a monthly alpha of 175 basis points. The signs and magnitudes of the alphas generally decrease monotonically from high to low ownership portfolios. In short, Table III documents a striking foreign ownership effect.
The observed foreign ownership effect is conditional on an assumed asset pricing model. Hence, an alternative interpretation of the evidence is that a conventional asset pricing model is inadequate in accounting for the systematic component of Taiwan stocks. In particular, the evidence calls for an additional risk factor, one related to foreign ownership. This foreign ownership factor may capture risks such as those associated with foreign exchange rates. Nonetheless, regardless of the interpretation, the evidence demonstrates that the foreign ownership level has an important pricing effect.
C. Robustness Checks
In this section, we conduct a series of robustness checks to determine whether the foreign ownership effect is due to some other effects. (18) First, we examine whether the foreign ownership effect is merely an export firm effect. It is not an export firm effect if the market rewards non-export-oriented firms with high foreign ownership levels. KS find that foreign ownership in Japan is concentrated in large export-oriented firms. They attribute this behavior to foreigners' attempts to mitigate their disadvantage in knowledge of domestic firms. This may be because it is more cost efficient for foreigners to track firms that have global operations. They may have easier access to information about those firms' customers, suppliers, and competitors. They may even have access to information in foreign markets that are unavailable to domestic investors.
To distinguish between the foreign ownership effect and the export firm effect, we sort stocks into three portfolios based on their export-to-sales ratios. Each export portfolio is then sorted into three ownership portfolios based on foreign ownership at the end of the previous quarter, resulting in nine export ownership portfolios.
Summary statistics indicate that foreign ownership is not synonymous with export ratios and that foreign ownership is not concentrated exclusively in export-oriented firms. Next, we examine the Jensen alphas of the nine export ownership portfolios. The results show that the foreign ownership effect is still present even after accounting for firm export ratios. Therefore, the foreign ownership effect is not an export firm effect. Apparently, investors acknowledge foreign ownership even in non-export-oriented Taiwanese firms. These are firms that are least likely to reflect any exchange rate effects.
We now investigate whether the foreign ownership effect is a size effect. As in the case of export-oriented firms, foreign owners may minimize their information disadvantage by investing in large firms. To investigate the size effect, we sort stocks into three size portfolios and then further sort each size portfolio into three ownership percentage portfolios for a total of nine portfolios.
Summary statistics confirm that foreign investment is concentrated in large stocks. However, controlling for size does not appear to control for foreign ownership. It also appears to be the case that controlling for foreign ownership does not control for foreign ownership across firm size. An examination of the Jensen alphas for both equal- and value-weighted versions of the nine size ownership portfolios indicates that the foreign ownership effect is not a size effect. The test results can also be interpreted as providing evidence regarding passive foreign institutional investments. A passive investment strategy for foreigners is to hold stocks in their Taiwan portfolio in the same proportion as that in an index. Such a passive strategy is also incompatible with the presence of a home bias as noted earlier. The popular Taiwan indices are value-weighted and are dominated by large firms. As such, our results may be viewed as being incompatible with a passive index fund strategy.
Foreign institutions may also minimize their information disadvantage by focusing their investments in transparent firms. Transparent firms release prompt and accurate disclosures and thereby reduce information asymmetry in the market. Because information is readily available about transparent firms, such firms should be attractive to investors who want to minimize their information asymmetry. Therefore, we ask whether the foreign ownership effect is a transparent firm effect.
We measure firm transparency by using the information transparency and disclosure ranking compiled by the Taiwan Securities and Futures Institute (2003) for all the listed stocks in Taiwan. Companies are ranked on five criteria: 1) compliance with mandatory disclosures, 2) timeliness of reporting, 3) disclosure of the annual report, 4) disclosure of a financial forecast, and 5) corporate website disclosure. Their 2003 ranking identifies firms that are considered to be "more transparent" companies. Again, we do a double sort with these data. We first sort stocks into five foreign ownership portfolios and then classify each foreign ownership portfolio into a firm that is either "more transparent" or "less transparent." This process of classification produces a total of 10 portfolios.
Summary statistics are consistent with foreigners' tendency to own larger firms that also happen to be more transparent. Jensen alphas for both equal- and value-weighted versions of the ten transparency ownership portfolios show a premium for "more transparent" firms. More importantly, they demonstrate that the foreign ownership effect is still present even after controlling for the firm transparency level. This suggests that foreign institutions do not minimize their information disadvantage by exclusively investing in transparent firms.
D. Discussion
Our results indicate that foreign ownership levels are associated with a pronounced local valuation effect. Stocks with high foreign ownership outperform stocks with low foreign ownership. The market rewards stocks that foreigners choose to own. Moreover, this reward is not simply because it is an export-oriented firm, a big firm, or a transparent firm.
Foreign investors are prohibited from short selling in Taiwan. This prohibition restricts foreign investors' potential to profit by identifying losers and may bias them toward strategies of ownership, oversight, and intervention. Although foreign investors are restricted from short selling, domestic investors are allowed to do so. Domestic investors can short sell to arbitrage away any perceived price abnormalities. However, any foreigners' buy and hold strategies because of their inability to short sell may restrict the supply of shares for short selling and may make arbitrage trading difficult. As such, part of the valuation effect may be due to the different short sale regulations. On the other hand, it is worth noting that if foreign ownership produces a positive long-term valuation effect, then foreign investors benefit from their ownership. Therefore, if foreign investors were permitted to engage in short sales, a positive valuation effect may be even stronger. This is because foreign investors would be able to short sell firms they do not want to own and to invest the proceeds in firms they do want to hold.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment