正在加载图片...
Since the control variables are highly persistent, we estimate their standard errors from the time-series of estimated coefficients via Newey-West [1987] and allow 12 monthly lags. Since media mentions within a publication are not persistent across months, and since for many of the media mentions we are unable o estimate a coefficient for each of the 60 months. we estimate the standard errors for the media mention variables via White[1980.(When we estimate standard errors for mentions in the Wall Street Journal and positive mentions in SmartMoney via Newey-West with 12 lags, they are quantitatively quite similar to those reported in Table V. Looking across the columns in Table V, we see that media mentions are associated with future fows in the direction we would expect. The magnitudes are largest in our baseline specification in column(2)and decline slightly when we control for fund is lagged Morningstar rating(column(3)), fund is prior media mentions(column(4), and the prior media mentions of other funds in fund is family(column(5)). Overall, ositive mentions in personal finance publications and Consumer Reports are associated with an economically significant 7-8 percent increase in fund size over the next 12 months, while a positive mention in the New York Times is associated with a 15 percent increase. Negative media mentions in both Kiplinger's and SmartMoney yield estimated future outflows of 1-2 percent, but neither coefficient is statistically significant The estimated coefficients on the control variables are largely consistent with those in the existing literature. For example, the positive and statistically significant coefficients on past fund returns and past fund returns squared capture the convex relation between inflows and performance first documented by Ippolito [1992. Interestingly, while fund-level inflows are increasing in non-print advertising expenditures they do not appear to be increasing in print advertising expenditures; in none of the columns is the coefficient on the print advertising ratio statistically different from zero. If we interpret the correlations in Tables Ill and V causally, this suggests that the returns to print advertising by mutual funds may be coming largel a biased content B. Do Media mentions Contain Information About Future returns? In this section, we compare the future returns of funds receiving media mentions to those of other funds The question of interest is whether investors can benefit from buying funds that receive positive mentionsSince the control variables are highly persistent, we estimate their standard errors from the time-series of estimated coefficients via Newey-West [1987] and allow 12 monthly lags. Since media mentions within a publication are not persistent across months, and since for many of the media mentions we are unable to estimate a coefficient for each of the 60 months, we estimate the standard errors for the media mention variables via White [1980]. (When we estimate standard errors for mentions in the Wall Street Journal and positive mentions in SmartMoney via Newey-West with 12 lags, they are quantitatively quite similar to those reported in Table V.) Looking across the columns in Table V, we see that media mentions are associated with future flows in the direction we would expect. The magnitudes are largest in our baseline specification in column (2) and decline slightly when we control for fund i’s lagged Morningstar rating (column (3)), fund i’s prior media mentions (column (4)), and the prior media mentions of other funds in fund i’s family (column (5)). Overall, positive mentions in personal finance publications and Consumer Reports are associated with an economically significant 7-8 percent increase in fund size over the next 12 months, while a positive mention in the New York Times is associated with a 15 percent increase. Negative media mentions in both Kiplinger’s and SmartMoney yield estimated future outflows of 1–2 percent, but neither coefficient is statistically significant. The estimated coefficients on the control variables are largely consistent with those in the existing literature. For example, the positive and statistically significant coefficients on past fund returns and past fund returns squared capture the convex relation between inflows and performance first documented by Ippolito [1992]. Interestingly, while fund-level inflows are increasing in non-print advertising expenditures, they do not appear to be increasing in print advertising expenditures; in none of the columns is the coefficient on the print advertising ratio statistically different from zero. If we interpret the correlations in Tables III and V causally, this suggests that the returns to print advertising by mutual funds may be coming largely via biased content. B. Do Media Mentions Contain Information About Future Returns? In this section, we compare the future returns of funds receiving media mentions to those of other funds. The question of interest is whether investors can benefit from buying funds that receive positive mentions 13
<<向上翻页向下翻页>>
©2008-现在 cucdc.com 高等教育资讯网 版权所有