Local effects of foreign ownership in an emerging financial market: evidence from qualified foreign institutional investors in Taiwan - III

Earlier research has focused on whether foreigner or domestic investors have a short-term information advantage and has yielded contradictory results. In contrast, we have focused on the long-term information advantage. The foreign ownership effect is obtained using monthly portfolios over a sample period of 90 months and the portfolios are rebalanced quarterly. Therefore, domestic investors have up to three months to react to news of a foreigner's transaction. Short-term results primarily reflect domestic investors' response to a foreign buy or sell that is most pronounced within the same trading day or over the following few days. Short-term evidence primarily reflects differential short lived advantages such as information about order flow imbalances. Long-term evidence above all primarily reflects differential long-lived advantages such as a firm's strategic information arising from technological, financial, or human expertise, experience, or resources. In short, short-term and long-term advantages are based on different information sets, and short-term results may or may not differ from long-term results.
The results provide evidence regarding the importance of foreign ownership in Taiwan and are compatible with casual empiricism. Individual investors in Taiwan are at a huge information disadvantage relative to firm management in an environment with extreme agency problems. Individual investors also do not have the resources to engage in stock research. Given a dearth of alternative mechanisms available for minimizing their information disadvantage, these investors may have focused on foreign ownership data. In response to the enthusiasm for foreign ownership, domestic firms have encouraged increased foreign ownership of their firms. The media have even reported homemade foreign investments, whereby Taiwanese individuals or investment entities buy Taiwan stocks through offshore accounts, giving the appearance of QFII buying.
We also observe foreign investor preferences similar to those first reported by KS for developed markets. Foreign investors may attempt to mitigate their information disadvantage by selecting firms that are export oriented, large, or transparent. However, our results are for an emerging financial market. More importantly, out results indicate that the foreign ownership effect extends to non-export-oriented, small, or nontransparent firms that are chosen by foreigners. This suggests that foreign ownership goes beyond export oriented, large, or transparent firms.

A natural follow up analysis is to address the reason why foreign ownership of local stocks is rewarded by the Taiwan market participants. What is the basis for domestic investors' focus on foreign ownership of local stocks? In the following section, we pursue a long-term performance based explanation.
IV. Monitoring and Foreign Ownership
The foreign ownership effect documented in the previous section suggests that if investors reward stocks that foreigners choose to own, foreigners must have an information advantage over domestic investors. With time, foreign investors will mitigate the information advantage domestic investors possess because of their familiarity with the local environment. However, what is it that foreigners know that domestic investors are unable to acquire in the long run? We hypothesize that it is the knowledge and the ability of foreign institutions to monitor their investments in Taiwan. Unlike domestic individual investors, foreign institutions have the expertise, experience, and resources to conduct firm research and to invest for the long term; they also have the reputation and credibility that neither domestic institutions nor individuals possess. This section provides evidence regarding whether the basis for investors' confidence in foreign ownership is rooted in foreign owners' monitoring ability.
In conducting our analysis, we are guided by previous research on firm monitoring. The existing literature is motivated by the agency problem that arises when management interests and those of firm owners diverge, resulting in excessive management perquisites. Jensen and Meckling (1976) propose increasing managerial ownership in order to harmonize managers' and shareholders' objectives, highlighting the role of monitoring by owners. Their insight has generated an extensive body of literature that examines the connection between firm ownership structure and firm performance. We consider a natural extension of this literature to our context by pursuing a performance based explanation of the foreign ownership effect. In Taiwan, principal-agent conflicts are potentially much more severe than those in developed markets.
A review of the monitoring literature suggests two major categories for examining the ownership-performance nexus. The first group of studies examines the link between ownership and performance in a specific firm activity. For example, Chen, Harford, and Li (2007) present evidence consistent with independent, long-term institutional investors who actively monitor and benefit from their monitoring activities. In particular, they provide evidence that firms with more independent long-term institutional investors have better postmerger performance when they bid and are more likely to make withdrawal from bad bids. Qiu (2006) finds that the presence of large public pension fund shareholders reduces ex ante bad acquisitions. Hartzell and Starks (2003) and Almazan, Hartzell, and Starks (2005) relate monitoring by institutional investors to executive compensation.
Although most of the papers in the first category consider mergers and acquisitions and executive compensation, we cannot examine either for Taiwan. During our sample period, mergers and acquisitions were rare events. As for executive compensation, the practice in Taiwanese firms is to keep the information hidden from the public. Not surprisingly, a lack of transparency translates into a lack of data availability. For our evidence, we choose R&D expenditures as firm activity. Bushee (1998) observes that there is an association between institutional monitoring and R&D expenditures.
To examine whether foreign institutional investors influence R&D expenditure decisions, we follow Bushee's (1998) approach in testing for the relation between foreign ownership and R&D expenditures. We collect annual R&D expenditures for our sample of firms from 1994 to 2001 and construct a binary dependent variable that is one if the firm increases R&D expenditures at year t and is zero otherwise. The regressors are annual foreign ownership and domestic institutional ownership variables for year t - 1 for years 1994-2000. We also include as control variables an industry indicator variable, a year indicator variable, and a proxy for a firm's profitability. The latter controls for increases in R&D spending that are associated with good times. We examine the results of three proxies: 1) change in return on assets, 2) change in EBITD (earnings before interest, taxes, and depreciations), and 3) the EBITD indicator variable that is one if EBITD increases and zero otherwise.
Table IV reports the results of the logistic regressions. Panel A demonstrates that the coefficient on foreign ownership is significantly positive, but the one on domestic institutional investors is negative. The results indicate that firms with higher foreign ownership are associated with increases in R&D expenditures. This provides evidence of a direct impact on managerial decisions of foreign investors' monitoring activity.
We conduct further analysis to examine whether the influence of monitoring on R&D expenditures is restricted to high-tech firms. Panels B and C illustrate that the relation is significant for both high-tech and non-high-tech firms, respectively. However, the relation seems to be stronger for non-high-tech firms. This may be because high-tech firms are better disciplined in investing in R&D.
The second set of studies in the earlier literature bypasses the business activities that improve firm performance and seeks to examine the relation between ownership structure and measures of firm performance. Mura (2007), Himmelberg, Hubbard, and Palia (1999), McConnell and Servaes (1990), and Morck, Shleifer, and Vishny (1988) measure firm performance by Tobin's Q and accounting returns. Similarly, we investigate firm improvement by using the same performance measures. Tobin's Q is the ratio of the sum of the market value of equity, the market value of preferred stock, and the book value of liabilities to the book value of total assets. We also use the book value of preferred stock in the numerator to estimate Tobin's Q and obtain similar results. (19) The accounting returns are for net income and EBITD.
In considering the alternative performance measures, we control for other effects that may impact the ownership-performance relation. We control for these effects by incorporating variables recommended in the literature: 1) standard deviation of returns, 2) debt ratio, 3) R&D expenses, and 4) sales (Smith and Watts, 1992; Anderson and Reeb, 2003). These firm characteristics are chosen to control for firm risk, leverage, size, and growth opportunity.
We also account for concentration of ownership in our analysis. Specifically, we include in our control variables a proxy for ownership concentration as measured by the percentage of shares held by insiders. We define insiders as officers of the firm and members of the board of directors. Our insider definition effectively includes all block holders in Taiwanese firms. Following the papers by McConnell and Servaes (1990) and Morck, Shleifer, and Vishny (1988), we permit a nonlinear impact of controlling shareholders in the Tobin's Q regressions by including both the proxy and its squared value. (20)
To estimate the ownership-performance relation, we regress firm performance on foreign ownership and firm characteristics. We address possible clustering effects in two ways. First, Himmelberg, Hubbard, and Palia (1999) observe that the regressions may result in biased estimates and spurious relations if the ownership variable is endogenous. Endogeneity arises when both ownership and performance are determined by common omitted variables, which may be unmeasurable. HHP recommend correcting for this problem with longitudinal data by using a fixed effects estimator under the assumption that omitted variables are fairly time invariant. (21) Foreign ownership may also be endogenous when performance affects ownership. In such situations, an instrumental variable can be used to estimate the relation. We use both the fixed effects and the instrumental variable estimators. To control for fixed effects, we use indicator variables to control for firm, year, and industry. (22) Although we only report the results with fixed effects for year and industry in the paper, similar results are obtained when fixed effects for firm are included as well. Of course, the latter has smaller degrees of freedom. Second, we use the Fama-MacBeth (1973) methodology to control for firm clustering in certain years.
Table V presents the fixed-effect results using annual panel data that take into account both heterogeneity across firms (as represented by industry indicator variables) and variation over time (as represented by year dummy variables). The Tobin's Q results are shown in Panel A and the accounting results are shown in Panel B. The regressions are conducted separately for contemporaneous and one-period lagged explanatory variables. We further examine each case both with and without the domestic institutional ownership variable. In all cases, foreign ownership variables are highly significant and positively associated with firm performance. In contrast, all domestic institutional variables are insignificantly different from zero. In the Tobin's Q regressions, the debt ratio coefficients are significantly negative and the R&D coefficients are significantly positive. Debt appears to be penalized, but R&D is valued by the market. The controlling ownership variable is positively and significantly related to foreign institutional ownership. Its association is also nonlinear in that its squared variable is negatively and significantly related to foreign ownership. In the accounting performance regressions, the debt ratio coefficients remain significantly negative, but contemporaneous R&D coefficients lose their significance. In addition, return standard deviation and sales coefficients are significantly negative and positive, respectively.
Table VI presents the instrumental variable results. In the first-stage regression, foreign ownership is regressed on market capitalization, the export ratio, and the one-period lagged firm performance measure. Size and export ratio are suggested by our earlier analyses, which demonstrate foreign investors' preference for large export-oriented firms. The last instrument exploits the autocorrelation in performance measures. In both the Tobin's Q and the accounting regressions, the coefficients on all three instruments are highly significantly positive. In the second-stage regressions, firm performance is regressed on the foreign ownership level predicted by the first-stage regression coefficients and other contemporaneous firm characteristics. The second-stage regressions are also run with and without the domestic institutional ownership variable. Again, all foreign ownership variables are highly significant in association with firm performance. The domestic institutional coefficients remain insignificant. As for the coefficients on firm characteristics, they are generally significant. In particular, the inside ownership coefficients maintain their significance and have the same sign in both Tables V and VI.
Table VII presents the Fama-MacBeth (1973) estimation results. They are similar to those in Tables V and VI. Most notably, foreign ownership is again positively and significantly associated with all three performance measures. In addition, domestic institutional ownership is either insignificant or, if significant, has a negative relation with the performance measure.