Google trends provides a time series search volume index (SVI) of search terms which are entered into Google search engine and helps us understand how popular a keyword is and realize the trends of the world. SVI is a relative index which is calculated from the frequency that the query term have been searched for each time period divided by the total number of searches done on Google search engine (Choi and Varian, 2012). Furthermore, the data is divided by the highest point and multiplied by 100 in the period selected by Google Trends users (Kim and Hanssens, 2011).
Therefore, the figures are displayed on a scale of 0 to 100 by the form of weekly SVI or monthly SVI.1 Google sets a threshold for search volume. When the search volume of a query term doesn’t pass the threshold, SVI will be presented by zero. Figure 1 shows the search results on Google Trends.
[Insert Figure 1 around here]
Traditionally, when researchers investigate the attention of investors to securities or products, they use advertisement expenditures, trading volume, newspaper headlines, and extreme returns (Da, Engelberg, and Gao, 2011) since the data of these variables reflect investor behaviors. However, the development of Internet alters people’s behavior. When people hear something attract them, they search on search engine to acquire more information. Over the past decade, with the progressive functions, Google search engine has the largest market shares in United States.2 The heavy search traffic of Google search engine drives Google trends Search Volume Index becomes a useful tool to predict present and reflects online buzz of products or securities (Choi and Varian, 2012). For example, Kim and Hanssens (2011) combine the SVI of Google Trends and Google Insight for Search to construct a new measure to proxy the online word-of-mouth and find that pre-launch advertising boosts the online buzz which positively influences the revenues of movies. Accordingly,
1 According to the introduction from Google Trends website.( http://www.google.com/trends/)
2 By November 2012, Google’s share of U.S. searches was 67%, according to comScore, which is an internet analytic information provider (http://www.comscore.com).
5
we infer that SVI may be useful when the searching objects are newly launched products such as newly issued funds since information about the new product from formal channels is few, people may search the new product on internet to obtain more related information when they are interested in the product (Da, et al., 2011). We also consider that SVI not only present the effect of advertisement and promotion but also present the effect of other factors such as word-of-mouth and online-buzz. We expect that SVI has good predictability on the cash flow of newly issued funds in the next period. Cash flows reflect the buying behavior of investors and affect the fund assets as well as the revenues of fund families. Therefore, when newly issued funds can attract more cash flows, they have higher survival probability.
Bank, Larch, and Peter (2011) investigate the effect of uninformed investor attention to stocks on stock liquidity and return of German stocks. They discover a positive relation between search volume of firm name and stock liquidity. Da et al. (2011, 2012) input the most popular product of 865 firms into Google insight to examine the predictability of SVI to revenues and earnings which surprise around earnings announcements and use SVI to form a new measure to predict the effect of investor sentiment on asset prices, volatility, and fund flow. They confirm that SVI is a serviceable measure to capture the demand side of product and information which represent the investors who have no direct and inside information about securities. Therefore, we will examine whether the SVI represent the “smart search” by sophisticated investors, which means that whether SVI can predict future returns.
Several studies verify that Google search volume index is a direct and leading proxy of uninformed investor attention (Bank et. al., 2011; Da et al., 2011; Goel et al., 2010). Unlike sophisticated investors who have more channels to access investment information easily, individual investors traditionally obtain information about mutual funds through external and lagged messages such as magazines, media, and financial consultants. However, since the progressively strong search function of internet reduces the searching cost and the information asymmetry among investors (Bank et al, 2011), uninformed investors can get information related to mutual funds by searching
6
these funds on internet. On the other hand, the decreasing cost of information also diminishes the risk to access investing information (Vlastakis and Markellos, 2012). When investors hear about something new about mutual funds which attract them, they will search fund names to obtain more information reducing perceived risk (Chang and Wu, 2012). According to Kim, Ferrin, and Rao (2007), perceived risk is defined as consumer’s recognition of the potential uncertain negative outcome they may encounter after products or securities transaction which affects the purchase decision. Perceived risk is an important factor which influences consumer’s purchase intention.
Several literatures find that reduced perceived risk increases purchase intention (Vijayasarathy and Jones, 2000; Park et al., 2005). Therefore, we expect an increase in Google search volume of fund is associated with an increase in subsequent fund cash flow and fund survival probability and develop the following hypothesis.
H1: Search volume index (SVI) has positive effect on fund cash flows and increase the survival probability of newly issued funds.
2.2 Fee Structure 2.2.1 Fund Expenses
Expense ratio which contains fund distribution and marketing fees, management fees, and operating costs is a form of expense that is a percentage of assets under management on annual basis. High expense ratios may worsen the adjusted returns of funds and influence cash flow (Rao, Boudreaux, and Das, 2012). Khorana and Servaes (2011) examine the effect of fund expense on fund family market shares and confirm that high expenses may harm for the market shares of fund families. Barber et al., 2005) use the 501 largest funds as sample and show low expense funds may have greater average market share since fund growth leads to lower expense. In other words, when funds achieve economies of scales, the operating cost can be amortized by the fund assets under management.
However, for new funds which have yet achieved economies of scales, the extreme competition between new funds and existed funds may induce fund managers pricing low expenses in order to
7
manifest highly fee adjusted returns and attract more cash flows. Therefore, we expect that funds with low expense ratio attract much more cash flows and have higher survival probability and develop the following hypothesis.
H2: Expense ratio has negative effect on fund cash flows and decreases the survival probability of newly issued funds.
2.2.2 Distribution and Marketing Fees
In the extent of marketing researches, there are plenty of studies corroborate that a product success is related to the firm’s marketing resource and skills, which adopt advertising, promotion, sales force and distribution, and product innovation (Panwar and Bapat, 2007). The execution of marketing activities increases the competitive advantages of exposure to target market and enhances the opportunity of success for new product (Calantone, Benedetto, and Song, 2011). This notion can be extend to fund industry. The most direct ways for mutual funds to examine the marketing activities are 12b-1 fees, front-end loads, and back-end loads which are paid to distribution and advertisement.
Funds often charge “load fees” such as front-end loads and back-end loads which are non-recurring charge. Front-end load is a distribution fee charged as investors buying a mutual fund, while back-end load is charged by the time of redemption. The charges of back-end loads are varied due to different redemption time of funds. Load fees are the most salient fees to investors compared to those annual fees which are calculated by a percentage of assets under management. Therefore, investors attend to loads more easily and try to avoid buying funds with high commission (Wilcox, 2003). Furthermore, Barber et al. (2005) verify a significant negative relation between front-end loads and new money of funds. For this reason, we contend that funds with high loads have higher failed probability than those with low commissions.
12b-1 fees, fees paid to distributors and known as one of marketing expenses, are contained in expense ratios. Advertisement can increase the fund awareness to investors and reduce searching
8
costs. Therefore, 12b-1 fees have positive effect on the market shares of fund families (Khorana and Servaes, 2011). Jain and Wu (2000) examine the effect of advertisement on fund flows and verify advertised funds indeed have significantly larger fund flows. Rao, Das, and Boudreaux (2012) show 12b-1 fee are successfully helping assets grow. Accordingly, we expect that high 12b-1 fees attract more cash flows to new funds, accumulate total net assets rapidly, and help the new funds survive.
H3: Load fees have negative effect on fund cash flows and decrease the survival probability of newly issued funds.
H4: 12b-1 fees have positive effect on fund cash flows and increase the survival probability of newly issued funds.
2.2.3 Multiple Class Shares
Fund industry is high homogeneity that leads to extreme competition of fund families. Thus product differentiation plays an important role of competition between mutual funds (Hortaçsu and Syverson, 2003). The main presentation of product differentiation for mutual funds is multiple class shares. Funds launch several classes with different fee structures to meet the preferences of investors. Funds with different classes have variety of fee structure. For instance, class A shares which charge front-end loads and lower 12b-1 fees are fit for long-term investors. Class B shares adapt to investors who want to invest for four to six years since they charge high back-end loads and 12b-1 fees that are adjusted depends on the time investors reimburse. Class C shares charge 1%
back-end loads and 1% 12b-1 fees for one year; therefore, the type of funds attracts short-term investors. As a result, as mutual funds contain multiple classes, they draw the attention of investors with different investment objectives, and attract more cash flows. Nanda, Wang, and Zheng (2009) verify this notion that multiple class funds grow more quickly than single class funds. Therefore, we expect the mutual funds with multiple classes contribute more cash inflow to their fund families and have greater survival possibility than single class funds.
H5: The funds class amount has positive effect on fund cash flows and increases the survival probability of newly issued funds.
9