Concluding Remarks

Limited market participation implies that institutional investors and financial analysts play diverse roles in information transmission in financial markets. To investigate the impact of limited market participation on the diverse roles played by institutional investors and financial analysts in propagating information, we form various portfolios that are different in the degree of institutional ownership as well as in the level of analyst coverage and conduct bivariate VAR analyses to measure the relative ability of one portfolio to predict the other.

We find that the ability of the portfolios with the highest institutional ownership to predict those with the lowest institutional ownership is better than vice versa, and that the ability of the portfolios with the highest analyst coverage to predict those with the lowest analyst coverage is better than vice versa. These imply that the firms that are primarily held by individual investors and followed by fewer financial analysts tend to respond more sluggishly to new market-wide information than do the firms that are primarily held by institutional investors and followed by more financial analysts.

More importantly, we find that within large size and high volume groups, the ability of the portfolios with the highest institutional ownership and those with the highest analyst coverage to predict the market portfolio is better than vice versa. However, within the small size and low volume groups, the ability of the market portfolio to predict the portfolios with the highest institutional ownership and those with the highest analyst coverage is better than vice versa. These findings demonstrate that because of the effect of limited market participation, institutional investors and financial analysts collect information more actively

about large and liquid stocks and, consequently, play dual roles of information discoverers and disseminators, whereas they do so less actively about small and illiquid stocks and, consequently, play dual roles of information receivers and disseminators.

To investigate the relative contributions of institutional investors and financial analysts in providing market-wide information to the market, we also construct analyst coverage-institutional ownership portfolios and institutional ownership-analyst coverage portfolios. We find that the ability of the portfolios with the highest institutional ownership to predict those with the lowest institutional ownership is better than vice versa for three of six analyst coverage groups, and that the ability of the portfolios with the highest analyst coverage to predict those with the lowest analyst coverage is better than vice versa for all of six institutional ownership groups. These findings imply that financial analysts provide more market-wide information to the market than do institutional investors.

As for the relative speed of the diffusion of good and bad market-wide news across securities, our results confirm the finding of McQueen et al. (1996) that good market-wide news travels more slowly across securities than does bad market-wide news. Moreover, we find that this asymmetric speed of the diffusion of good and bad market-wide news across securities primarily occurs during the periods of NBER-dated expansions.

One interesting finding obtained from our empirical analyses is that the returns of some specific portfolios such as the portfolios with larger market capitalization and the highest institutional ownership lead the returns on the market portfolio. However, it is not clear whether investors could profit from observing the price movements of stocks with these characteristics to devise their trading strategies once transaction costs are taken into account. A more rigorous study of this issue is warranted for future research.

References

Akaike, H., 1974. A new look at the statistical identification model, IEEE Trans. Auto.

Control, 19, 716-723.

Badrinath, S. G., Kale, J. R., Noe, T. H., 1995. Of shepherds, sheep and the cross-autocorrelations in equity returns. Review of Financial Studies 8, 401-430.

Barclay, M. J., Litzenberger, R. H., Warner, J. B., 1990. Private information, trading volume, and stock-return variances. Review of Financial Studies 3, 233-253.

Bennett, J. A., Sias, R. W., Starks, L. T., 2003. Greener pastures and the impact of dynamic institutional preferences. Review of Financial Studies 16, 1203-1238.

Boudoukh, J., Richardson, M., Whitelaw, R. F., 1994. A tale of three schools: Insights on autocorrelations of short-horizon stock returns. Review of Financial Studies 7, 539-573.

Brennan, M. J., Jegadeesh, N., Swaminathan, B., 1993. Investment analysis and the adjustment of stock prices to common information. Review of Financial Studies 6, 799-824.

Chordia, T., Swaminathan, B., 2000. Trading volume and cross-autocorrelations in stock returns. Journal of Finance 55, 913-935.

Conrad, J., Kaul, G., 1988. Time varying expected returns. Journal of Business 61, 409-425.

Foster, F. D., Viswanathan, S., 1993. The effect of public information and competition on trading volume and price volatility. Review of Financial Studies 6, 23-56.

Holden, C., Subrahmanyam, A., 1992. Long-lived price information and imperfect competition. Journal of Finance 47, 247-270.

Hong, H., Stein, J. C., 1999. A united theory of underreaction, momentum trading, and overreaction in asset markets. Journal of Finance 54, 2143-2184.

Hong, H., Lim, T., Stein, J. C., 2000. Bad news travels slowly: Size, analyst coverage, and the profitability of momentum strategies. Journal of Finance 55, 265-296.

Kadlec, G. B., Patterson, D. M., 1999. A transactions data analysis of nonsynchronous

trading. Review of Financial Studies 12, 609-630.

Kyle, A. S., 1985. Continuous auctions and insider trading. Econometrica 53, 1315-1335.

Lo, A. W., Mackinlay, A. C., 1990. When are contrarian profits due to stock market overreaction. Review of Financial Studies 3, 175-206.

Lo, A. W., Wang, J., 2000. Trading volume: Definition, data analysis, and implications of portfolio theory. Review of Financial Studies 13, 257-300.

McQueen, G., Pinegar, M., Thorley, S., 1996. Delayed reaction to good news and the cross-autocorrelation of portfolio returns. Journal of Finance 51, 889-919.

Merton, R. C., 1987. A simple model of capital market equilibrium with incomplete information. Journal of Finance 42, 483-510.

Nofsinger, J. R., Sias, R. W., 1999. Herding and feedback trading by institutional and individual investors. Journal of Finance 54, 2263-2295.

Schwarz, G., 1978, Estimating the dimension of a model, Annals of Statistics 6, 461-464.

Sias, R. W., Starks, L. T., 1997. Return autocorrelation and institutional investors. Journal

of Financial Economics 46, 103-131.

Table 1

Summary Statistics for Various Portfolios

Summary statistics for various portfolios are computed for the sample period from January 1983 to December 2004. For size groups, Pij refers to a portfolio of size i and institutional-ownership or analyst coverage j. i = 1, 6 refer to the largest and smallest size portfolios, respectively. h and l refer to the highest and lowest institutional-ownership or analyst coverage portfolios, respectively, within each size group i. For volume groups, Pij refers to a portfolio of volume i and institutional-ownership or analyst coverage j. i = 1, 6 refer to the highest and lowest volume portfolios, respectively. h and l refer to the highest and lowest institutional-ownership or analyst coverage portfolios, respectively, within each volume group i. For analyst coverage groups, Pij refers to a portfolio of analyst coverage i and institutional ownership j. i = 1, 6 refer to the highest and lowest analyst coverage portfolios, respectively. h and l refer to the highest and lowest institutional ownership portfolios, respectively, within each analyst coverage group i. For institutional ownership groups, Pij refers to a portfolio of institutional ownership i and analyst coverage j. i = 1, 6 refer to the highest and lowest institutional ownership portfolios, respectively. h and l refer to the highest and lowest analyst coverage portfolios, respectively, within each institutional ownership group i. Pem and Pvm refer to the equal- and value-weighted portfolios of all NYSE sample firms, respectively. ρ₁ refers to the first-order autocorrelation.

S10 refers to the sum of the first 10 autocorrelations. The mean size figures are in billions of dollars. The mean volume figures represent the average percentage of trading turnover. The mean institutional ownership figures are in institutional ownership fraction. The mean analyst coverage figures represent the average number of analysts following a sample firm. The t-statistics for Pi vs. Pi+1 are the result of t-test for the difference in means of group i and group i + 1 of portfolio formation criterion 1. For example, for the size-institutional ownership portfolios, portfolio formation criterion 1 represents the mean market capitalization. The t-statistics for Pih vs. Pil are the results of t-test for the difference in means of groups h and l within each group i of portfolio formation criterion 2. For example, for the size-institutional ownership portfolios, portfolio formation criterion 2 represents the mean institutional ownership fraction.

Portfolio Returns Mean

(%)

Std. Dev.

(%)

Size Volume (%)

Institutional Ownership

Analyst Coverage

t-statistics for Pi vs. Pi+1

t-statistics for Pih vs. Pil

Size-Institutional Ownership Portfolios

P1_h 0.061 1.101 9.864 0.416 0.806 21.278

P1_l 0.064 0.857 18.952 0.263 0.324 22.776

11.665*** 138.306***

P2_h 0.061 1.111 2.742 0.482 0.829 17.006

P2_l 0.051 0.827 2.607 0.270 0.281 16.605

17.014*** 80.667***

P3_h 0.064 1.091 1.228 0.511 0.821 13.991

P3_l 0.056 0.821 1.140 0.270 0.225 11.971

20.798*** 91.851***

P4_h 0.061 1.085 0.624 0.490 0.805 11.090

P4_l 0.062 0.855 0.602 0.235 0.190 7.779

17.307*** 121.557***

P5_h 0.057 1.101 0.318 0.435 0.774 8.385

P5_l 0.061 0.934 0.291 0.250 0.171 5.370

19.225*** 91.184***

P6_h 0.072 1.131 0.132 0.424 0.661 6.094

P6_l 0.105 1.271 0.079 0.265 0.114 3.385

— 51.223***

Volume-Institutional Ownership Portfolios

P1_h 0.047 1.301 2.250 0.919 0.838 15.721

P1_l 0.163 1.481 1.325 0.950 0.270 8.444

26.779*** 46.898***

P2_h 0.072 1.067 2.679 0.466 0.819 14.381 28.954*** 70.534***

P2_l 0.079 1.254 2.332 0.459 0.282 9.322 P3_h 0.076 0.991 3.732 0.340 0.800 14.344

P3_l 0.081 1.110 2.651 0.334 0.269 9.196

28.945*** 65.898***

P4_h 0.064 0.926 3.971 0.260 0.774 13.436

P4_l 0.074 1.031 2.165 0.254 0.243 9.523

27.284*** 70.022***

P5_h 0.063 0.913 3.849 0.193 0.730 12.808

P5_l 0.056 0.946 1.879 0.185 0.195 8.670

23.033*** 81.477***

P6_h 0.053 0.816 2.402 0.115 0.656 9.563

P6_l 0.049 0.880 1.222 0.082 0.114 5.297

— 35.926***

Size-Analyst Coverage Portfolios

P1_h 0.048 1.084 24.043 0.383 0.582 33.432

P1_l 0.065 1.157 9.635 0.403 0.604 12.981

11.505*** 66.736***

P2_h 0.049 1.055 3.619 0.504 0.590 25.218

P2_l 0.062 1.069 3.052 0.381 0.570 9.110

17.596*** 57.912***

P3_h 0.042 0.971 1.708 0.586 0.586 21.888

P3_l 0.074 1.102 1.472 0.342 0.507 5.843

23.510*** 53.728***

P4_h 0.066 1.212 0.870 0.560 0.581 17.461

P4_l 0.047 1.000 0.753 0.360 0.512 3.557

25.221*** 45.182***

P5_h 0.079 1.231 0.458 0.509 0.560 13.350

P5_l 0.047 0.982 0.382 0.296 0.403 2.043

22.921*** 36.051***

P6_h 0.083 1.299 0.198 0.530 0.498 9.504

P6_l 0.081 1.268 0.107 0.324 0.329 1.145

— 24.921***

Volume-Analyst Coverage Portfolios

P1_h 0.070 1.385 6.046 0.957 0.658 31.245 5.840*** 45.505***

P1_l 0.084 1.500 0.603 0.811 0.460 3.188

P2_h 0.062 1.135 10.193 0.443 0.636 26.660 28.751*** 64.244***

P2_l 0.061 1.241 0.464 0.436 0.480 3.500

P3_h 0.058 1.031 14.605 0.329 0.603 27.318 21.676*** 59.483***

P3_l 0.079 1.117 0.486 0.326 0.461 3.253

P4_h 0.071 1.010 16.991 0.258 0.561 26.955 28.512*** 63.209***

P4_l 0.067 1.080 0.496 0.253 0.454 2.962

P5_h 0.057 0.984 22.026 0.196 0.498 25.877 21.174*** 73.806***

P5_l 0.050 1.022 1.061 0.190 0.407 2.452

P6_h 0.060 0.880 21.841 0.134 0.431 20.869 — 45.233***

P6_l 0.049 0.845 0.398 0.098 0.291 1.517 Analyst Coverage-Institutional Ownership Portfolios

P1_h 0.063 1.131 7.666 0.533 0.793 25.523

P1_l 0.059 0.946 14.969 0.284 0.346 24.990

47.702*** 66.537***

P2_h 0.060 1.119 3.680 0.511 0.814 17.915

P2_l 0.058 0.885 3.755 0.303 0.308 17.853

63.259*** 123.952***

P3_h 0.061 1.133 2.307 0.496 0.814 13.036

P3_l 0.067 0.965 1.677 0.294 0.285 12.837

36.823*** 96.001***

P4_h 0.057 1.101 1.368 0.432 0.805 9.206

P4_l 0.063 0.960 1.017 0.294 0.225 8.899

47.652*** 138.193***

P5_h 0.065 1.061 0.823 0.396 0.780 5.872

P5_l 0.079 0.989 0.608 0.243 0.180 5.522

35.612*** 95.067***

P6_h 0.048 1.104 0.425 0.345 0.684 2.725

P6_l 0.080 1.175 0.390 0.207 0.133 2.172

— 60.464***

Institutional Ownership-Analyst Coverage Portfolios

P1_h 0.066 1.127 7.578 0.530 0.785 26.370

P1_l 0.057 1.061 1.009 0.464 0.798 5.232

70.775*** 63.409***

P2_h 0.067 1.145 10.035 0.452 0.678 27.726

P2_l 0.049 1.103 0.565 0.351 0.671 4.682

86.180*** 71.147***

P3_h 0.069 1.122 17.235 0.360 0.588 26.869

P3_l 0.047 1.027 0.531 0.327 0.584 3.696

91.225*** 68.246***

P4_h 0.063 1.051 28.389 0.315 0.496 27.653

P4_l 0.061 1.057 0.513 0.316 0.487 2.855

92.139*** 77.440***

P5_h 0.049 0.980 19.592 0.283 0.386 24.568

P5_l 0.058 1.082 0.356 0.301 0.370 2.075

52.658*** 73.214***

P6_h 0.061 0.962 5.588 0.305 0.229 18.789

P6_l 0.081 1.197 0.346 0.231 0.176 1.366

— 50.942***

Market Portfolios

P_em 0.062 0.741 — — — — — —

P_vm 0.054 0.938 — — — — — —

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

Table 2

Vector Autoregressions for the Size-Institutional Ownership and the Market Portfolios

The following bivariate VAR is estimated to examine the relative ability of one portfolio to predict the other portfolio for the sample period from January 1983 to December 2004:

, , , , equal-weighted portfolio of size i and institutional-ownership j. i = 1, 6 refer to the largest and smallest size portfolios, respectively. h and l refer to the highest and lowest institutional-ownership portfolios, respectively, within each size group i. Pem and Pvm refer to the equal- and value-weighted portfolios of all NYSE sample firms, respectively. The number of lags in each equation is chosen by considering both the Akaike (1974) information criterion (AIC) and the Schwarz (1978) information criterion (SIC). The χ_b( )K and χ_c( )K statistics obtained from the Wald test are a joint test of the null hypothesis based on the causality restrictions.

The χ_b(1) and χ_c(1) statistics obtained from the Wald test are used to test the null hypothesis that

3 ,h t

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

Table 3

Vector Autoregressions for the Volume-Institutional Ownership and the Market Portfolios

The following bivariate VAR is estimated to examine the relative ability of one portfolio to predict the other portfolio for the sample period from January 1983 to December 2004:

, , , , equal-weighted portfolio of volume i and institutional-ownership j. i = 1, 6 refer to the highest and lowest volume portfolios, respectively. h and l refer to the highest and lowest institutional-ownership portfolios, respectively, within each volume group i. Pem and Pvm refer to the equal- and value-weighted portfolios of all NYSE sample firms, respectively. The number of lags in each equation is chosen by considering both the Akaike (1974) information criterion (AIC) and the Schwarz (1978) information criterion (SIC). The χ_b( )K and χ_c( )K statistics obtained from the Wald test are a joint test of the null hypothesis based on the causality restrictions. The χ_b(1) and χ_c(1) statistics obtained from the Wald test are used to test the null hypothesis that

∑

b_k =0 ^{and that}

∑

^{ck =0,} respectively. The χ_bc(1) statistic obtained from the Wald test is used

,

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

Table 4

Vector Autoregressions for the Size-Analyst Coverage and the Market Portfolios

The following bivariate VAR is estimated to examine the relative ability of one portfolio to predict the other portfolio for the sample period from January 1983 to December 2004:

, , , , equal-weighted portfolio of size i and analyst coverage j. i = 1, 6 refer to the largest and lowest size portfolios, respectively. h and l refer to the highest and lowest analyst coverage portfolios, respectively, within each size group i. Pem and Pvm refer to the equal- and value-weighted portfolios of all NYSE sample firms, respectively.

The number of lags in each equation is chosen by considering both the Akaike (1974) information criterion (AIC) and the Schwarz (1978) information criterion (SIC). The χ_b( )K and χ_c( )K statistics obtained from the Wald test are a joint test of the null hypothesis based on the causality restrictions. The χ_b(1) and χ_c(1) statistics obtained from the Wald test are used to test the null hypothesis that

∑

b_k =0 ^{and that}

∑

^{ck =0,}

respectively. The χ_bc(1) statistic obtained from the Wald test is used to test the null hypothesis that

k k.

,

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

Table 5

Vector Autoregressions for the Volume-Analyst Coverage and the Market Portfolios

The following bivariate VAR is estimated to examine the relative ability of one portfolio to predict the other portfolio for the sample period from January 1983 to December 2004:

, , , , equal-weighted portfolio of volume i and analyst coverage j. i = 1, 6 refer to the highest and lowest volume portfolios, respectively. h and l refer to the highest and lowest analyst coverage portfolios, respectively, within each volume group i. Pem and Pvm refer to the equal- and value-weighted portfolios of all NYSE sample firms, respectively. The number of lags in each equation is chosen by considering both the Akaike (1974) information criterion (AIC) and the Schwarz (1978) information criterion (SIC). The χ_b( )K and χ_c( )K statistics obtained from the Wald test are a joint test of the null hypothesis based on the causality restrictions.

The χ_b(1) and χ_c(1) statistics obtained from the Wald test are used to test the null hypothesis that

,

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

Table 6

Vector Autoregressions for the Analyst Coverage-Institutional Ownership and the Market Portfolios

The following bivariate VAR is estimated to examine the relative ability of one portfolio to predict the other portfolio for the sample period from January 1983 to December 2004:

, , , , equal-weighted portfolio of analyst coverage i and institutional ownership j. i = 1, 6 refer to the highest and lowest analyst coverage portfolios, respectively. h and l refer to the highest and lowest institutional ownership portfolios, respectively, within each analyst coverage group i. Pem and Pvm refer to the equal- and value-weighted portfolios of all NYSE sample firms, respectively. The number of lags in each equation is chosen by considering both the Akaike (1974) information criterion (AIC) and the Schwarz (1978) information criterion (SIC). The χ_b( )K and χ_c( )K statistics obtained from the Wald test are a joint test of the null hypothesis based on the causality restrictions. The χ_b(1) and χ_c(1) statistics obtained from the Wald test are used to test the null hypothesis that

∑

b_k =0 ^{and that}

∑

^{ck =0,} respectively. The χ_bc(1) statistic obtained from the Wald test is used to test the null hypothesis that

∑

b_k =

∑

c_k.

Panel A: Pih (Portfolio A) vs. Pil (Portfolio B)

,

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

Table 7

Vector Autoregressions for the Institutional Ownership-Analyst Coverage and the Market Portfolios

The following bivariate VAR is estimated to examine the relative ability of one portfolio to predict the other portfolio for the sample period from January 1983 to December 2004:

, , , , equal-weighted portfolio of institutional ownership i and analyst coverage j. i = 1, 6 refer to the highest and lowest institutional ownership portfolios, respectively. h and l refer to the highest and lowest analyst coverage portfolios, respectively, within each institutional ownership group i. Pem and Pvm refer to the equal- and value-weighted portfolios of all NYSE sample firms, respectively. The number of lags in each equation is chosen by considering both the Akaike (1974) information criterion (AIC) and the Schwarz (1978) information criterion (SIC). The χ_b( )K and χ_c( )K statistics obtained from the Wald test are a joint test of the null hypothesis based on the causality restrictions. The χ_b(1) and χ_c(1) statistics obtained from the Wald test are used to test the null hypothesis that

∑

b_k =0 ^{and that}

∑

^{ck =0,} respectively. The χ_bc(1) statistic obtained from the Wald test is used to test the null hypothesis that

∑

b_k =

∑

c_k.

Panel A: Pih (Portfolio A) vs. Pil (Portfolio B)

,

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

Table 8

Asymmetric Regression Based on the Sign of Portfolio Returns

The following regression is estimated to examine the asymmetric response of the returns of one portfolio to positive and negative returns of the other portfolio for the sample period from January 1983 to December 2004: equal-weighted portfolio of size i and institutional-ownership or analyst coverage j. i = 1, 6 refer to the largest and smallest size portfolios, respectively. h and l refer to the highest and lowest institutional-ownership or analyst coverage portfolios, respectively, within each size group i. For volume groups, Pij refers to an equal-weighted portfolio of volume i and institutional-ownership or analyst coverage j. i = 1, 6 refer to the highest and lowest volume portfolios, respectively. h and l refer to the highest and lowest institutional-ownership or analyst coverage portfolios, respectively, within each volume group i. The number of lags in the equation is chosen by considering both the Akaike (1974) information criterion (AIC) and the Schwarz (1978) information criterion (SIC). The t-statistics are reported in parentheses and are corrected for autocorrelations and heteroskedasticity using the Newey-West (1987) covariance matrix. The χ²(1) test statistic is used to test the null hypothesis that

∑

^K_k=1β_ik^UP=0 ^{and that}

∑

^K_k=1β_ik^DN =0. ^{The χ12(1) test} statistic is used to test the null hypothesis that β_i^UP₀ =β_i^DN₀ . The χ₂²(1) test statistic is used to test the null hypothesis that

∑

^K_k=1β_ik^UP =

∑

^K_k=1β_ik^DN. ^R2 is the adjusted coefficient of determinant.

Panel A: Size-Institutional Ownership Portfolios (Pih = Portfolio A and Pil = Portfolio B)

LHS variable R_{1 ,l t} R_{2 ,l t} R_{3 ,l t} R_{4 ,l t} R_{5 ,l t} R_{6 ,l t}

∑

0.077 0.058 0.074 0.040 0.068 0.285

2(1)

Panel B: Volume-Institutional Ownership Portfolios (Pih = Portfolio A and Pil = Portfolio B)

LHS variable R_{1 ,l t} R_{2 ,l t} R_{3 ,l t} R_{4 ,l t} R_{5 ,l t} R_{6 ,l t}

2

∑

0.278 0.304 0.133 0.224 0.168 0.265

2(1)

Panel C: Size-Analyst Coverage Portfolios (Pih = Portfolio A and Pil = Portfolio B)

LHS variable R_{1 ,l t} R_{2 ,l t} R_{3 ,l t} R_{4 ,l t} R_{5 ,l t} R_{6 ,l t}

∑

0.011 0.011 0.105 0.038 0.035 0.161

2(1)

∑

-0.210 -0.016 -0.121 0.099 0.179 0.065

2(1)

Panel D: Volume-Analyst Coverage Portfolios (Pih = Portfolio A and Pil = Portfolio B)

LHS variable R_{1 ,l t} R_{2 ,l t} R_{3 ,l t} R_{4 ,l t} R_{5 ,l t} R_{6 ,l t}

∑

0.168 0.101 0.071 0.234 0.144 0.165

2(1)

∑

0.145 0.225 0.440 1.105 0.181 0.195

2(1)

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

Table 9

Asymmetric Regression Based on the Sign of Portfolio Returns and the State of the Macroeconomy

The following regression is estimated to examine the asymmetric response of the returns of one portfolio to positive and negative returns of the other portfolio conditional on the state of the macroeconomy for the sample period from January 1983 to December 2004:

UP-EXP EXP DN-EXP EXP

, , , , , , ,

0 0

UP-CON EXP DN-CON EXP

, , , , , , , on a value of one during an NBER-dated expansion and zero otherwise. For size groups, Pij refers to an equal-weighted portfolio of size i and institutional-ownership or analyst coverage j. i = 1, 6 refer to the largest and smallest size portfolios, respectively. h and l refer to the highest and lowest institutional-ownership or analyst coverage portfolios, respectively, within each size group i. For volume groups, Pij refers to an equal-weighted portfolio of volume i and institutional-ownership or analyst coverage j. i = 1, 6 refer to the highest and lowest volume portfolios, respectively. h and l refer to the highest and lowest institutional-ownership or analyst coverage portfolios, respectively, within each volume group i. The number of lags in the equation is chosen by considering both the Akaike (1974) information criterion (AIC) and the Schwarz (1978) information criterion (SIC). The t-statistics are reported in parentheses and are corrected for autocorrelations and heteroskedasticity using the Newey-West (1987) covariance matrix. The χ²(1) test statistic is used to test the null hypothesis that the sum of the lagged coefficients is equal to zero.The χ₁²(1) test statistic is used to test the null hypothesis that β_i^UP-EXP₀ =β_i^DN-EXP₀ and that β_i^UP-CON₀ =β_i^DN-CON₀ . The χ₂²(1)

∑ ∑

^R2 is the adjusted coefficient of determinant.

Panel A: Size-Institutional Ownership Portfolios (Pih = Portfolio A and Pil = Portfolio B)

LHS variable R_{1 ,l t} R_{2 ,l t} R_{3 ,l t} R_{4 ,l t} R_{5 ,l t} R_{6 ,l t}

∑

0.050 0.051 0.066 0.035 0.010 0.369

2(1)

2

∑

-0.053 0.025 -0.011 -0.028 -0.033 0.049

2(1)

Panel B: Volume-Institutional Ownership Portfolios (Pih = Portfolio A and Pil = Portfolio B)

LHS variable R_{1 ,l t} R_{2 ,l t} R_{3 ,l t} R_{4 ,l t} R_{5 ,l t} R_{6 ,l t}

∑

0.264 0.316 0.127 0.136 0.076 0.074

2(1)

∑

0.058 0.136 0.058 0.169 0.142 0.142

2(1)

2

Panel C: Size-Analyst Coverage Portfolios (Pih = Portfolio A and Pil = Portfolio B)

LHS variable R_{1 ,l t} R_{2 ,l t} R_{3 ,l t} R_{4 ,l t} R_{5 ,l t} R_{6 ,l t}

∑

0.007 -0.054 0.105 0.067 0.089 0.116

2(1)

∑

-0.244 -0.057 -0.111 0.136 0.258 0.070

2(1)

∑

-0.028 -0.014 0.013 -0.023 -0.051 0.203

2(1)

∑

-0.015 0.103 0.040 0.008 0.014 0.156

2(1)

Panel D: Volume-Analyst Coverage Portfolios (Pih = Portfolio A and Pil = Portfolio B)

LHS variable R_{1 ,l t} R_{2 ,l t} R_{3 ,l t} R_{4 ,l t} R_{5 ,l t} R_{6 ,l t}

∑

0.200 0.120 0.078 0.235 0.167 0.079

2(1)

DN-EXP 1 , K k₌ βil k

∑

0.184 0.266 0.475 0.114 0.181 0.165

2(1)

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

研討會心得報告

此次本人參加 2007 年財務管理協會的年會，是發表 95 年度之國科會計畫。當時 為了趕上投稿日期，真是忙的不可開交，很幸運地還是來得及完成部份研究成果 去投稿。財務管理協會的年會是全球規模最大的財務研討會，聽說投稿通過率一 年比一年低。此次文章的評論人提出了一些很有建設性的評論，有助於文章的進 一步修改。此外，自己也參加了幾場自己對相關主題有興趣的場次，發覺可從中 找到一些未來的研究靈感。因此，本人覺得參加此次財務管理協會年會，收穫頗 豐。在此，對國科會有一項建議，國科會網站上所公告的飛機票票價，跟真實票 價有一段不小的差距，希望未來在飛機票票價的訂定上，能較符合市場的真實價

此次本人參加 2007 年財務管理協會的年會，是發表 95 年度之國科會計畫。當時 為了趕上投稿日期，真是忙的不可開交，很幸運地還是來得及完成部份研究成果 去投稿。財務管理協會的年會是全球規模最大的財務研討會，聽說投稿通過率一 年比一年低。此次文章的評論人提出了一些很有建設性的評論，有助於文章的進 一步修改。此外，自己也參加了幾場自己對相關主題有興趣的場次，發覺可從中 找到一些未來的研究靈感。因此，本人覺得參加此次財務管理協會年會，收穫頗 豐。在此，對國科會有一項建議，國科會網站上所公告的飛機票票價，跟真實票 價有一段不小的差距，希望未來在飛機票票價的訂定上，能較符合市場的真實價

在文檔中 機構投資人所扮演之訊息傳遞角色 (頁 24-49)