Description
In finance, liquidity risk is the risk that a given security or asset cannot be traded quickly enough in the market to prevent a loss (or make the required profit).
Study of Advertising on Liquidity Risk
Abstract Marketing actions can impact firm value through both product and capital markets. Recent literature in finance suggests that in addition to systematic market risk, liquidity risk the non-diversifiable systematic risk that a firm's stock may become illiquid in times of market stress - is also priced. The authors examine a large panel of more than 1,800 firms from 1971 to 2005 and show that advertising lowers liquidity risk by increasing the number of individual investors in a firm. The impact of higher advertising on liquidity risk is more pronounced for firms that are younger, operate in BtoC markets, and have advertising expenditures between about $ .5 million and $ 20 million. Simulations show that increasing advertising expenditures by 25% increases firm value by up to 1.4% just due to a reduction in liquidity risk. Overall, the results suggest that, in addition to the product market impacts, researchers and managers should consider the valuation impact of marketing activities via their effect on investors in capital markets. Key words: Advertising, Shareholder value, Liquidity risk
1
The marketing profession has demonstrated the positive effect of marketing actions on shareholder value, either by improving future cash flows or by reducing their riskiness (Srinivasan and Hanssens 2009). Several earlier studies have examined the influence of marketing on a firm's cash flow (e.g., Mizik and Jacobson 2007, Fornell et. al 2006). More recently researchers have studied the impact of marketing activities on firm risk, using various risk metrics. A partial list includes the effect of advertising (McAlister, Srinivasan and Kim 2007) and customer satisfaction (Tuli and Bharadwaj 2009) on systematic market risk; consumer negative voice (Luo 2007), CSR actions (Luo and Bhattacharya 2009), and innovation (Sorescu and Spanjol 2008) on idiosyncratic firm risk; customer satisfaction (Gruca and Rego 2005) on cash flow variability; and customer relationships on sales volatility (Tuli, Bharadwaj, Kohli 2010). Investigating the impact of marketing activities on risk is clearly an important topic for marketers to study. In this paper, we examine whether advertising by a firm impacts liquidity risk, a systematic risk that has not been explored in the extant marketing literature. Recent literature in finance suggests that in addition to the commonly understood systematic market risk, investors are also concerned about systematic liquidity risk. Liquidity represents the ease with which an investor is able to trade a stock at the market price, at low cost, in sufficient quantities quickly. Liquidity varies over time both for individual stocks and for the market as a whole. Liquidity risk represents the non-diversifiable systematic risk that a firm's stock may become illiquid in times of market stress. For example, during market downturns when the market declines in value or overall liquidity dries up, investors may be unable to sell some of their assets quickly to meet consumption or other needs. Hence, investors would be willing to pay a premium and accept lower returns on a stock that is expected to be liquid in down or illiquid markets due to its low covariance with the market. If a stock has low liquidity
2
risk, investors require a smaller risk premium to compensate for this systematic risk, thus decreasing the firm's cost of capital. This decrease in cost of capital will increase firm value, all else equal.i Acharya and Pedersen (2005) find that the cost of equity is lower by about 1.1 % per year for stocks with low liquidity risk. Liquidity risk is shown to be priced not only in the US but also across a large number of global markets, thereby suggesting that it is a pervasive risk which is important in determining firm value (Lee 2010). In this paper, we explore the role of marketing, and more specifically advertising, in reducing liquidity risk through its impact on investor behavior in capital markets. We draw upon the prior literature in finance and hypothesize that a firm may be able to reduce its liquidity risk by increasing the number of individual investors. Both theoretical and empirical results in the finance literature suggest that improved awareness can increase the number of investors. Merton (1987) asserts that investors are likely to invest only in firms they know about. Consistent with this view, Huberman (2001) finds that investors are more likely to hold shares of local firms that they are aware of, and concludes that "people invest in the familiar while often ignoring the principles of portfolio theory.? Since advertising is one of the main instruments available to marketing managers to increase awareness, we examine its role in reducing liquidity risk. Anecdotal evidence suggests that managers realize the importance of improving individual investor awareness. For example, Federated Department Stores changed its corporate name to Macy's, and stated that ?by aligning our corporate name with our largest brand, we will increase the visibility of the company with customers, leverage the world-famous Macy's brand name, and get more credit for our accomplishments in the marketplace? (Lundgren 2007; italics added). Similarly, Sun Microsystems changed its trading symbol from SUNW to
3
JAVA to make the firm more visible to new investors (Jonathan Schwartz, President and CEO, 2007). Our analysis proceeds along four related lines of inquiry. First, we compare the liquidity risk of firms that do not advertise with firms that report advertising expenses. Using data on more than 1,800 publicly traded US firms over a thirty-five year period (1971 to 2005); we find that firms with no advertising are subject to significantly higher liquidity risk than firms that advertise. We obtain qualitatively similar results when we use size-matched firms, suggesting that our results are not an artifact of firm size. Second, we investigate whether the level of advertising affects the level of liquidity risk. We document that liquidity risk declines monotonically across quartiles when firms are grouped annually based on the reported advertising expense. The results are qualitatively similar when we use alternative model specifications; therefore we conclude that the level of advertising is highly correlated with the level of liquidity risk. Third, we show that a firm's liquidity risk is negatively related to the number of individual investors. We also show that advertising is more effective in reducing liquidity risk in BtoC (compared to BtoB) firms as well as in younger firms. Taken together, these results support the notion that product advertising improves a firm's awareness among individual investors in capital markets and lowers liquidity risk. Finally, we assess the economic significance of this result by simulating the valuation impact of potentially increasing advertising expenditures by 25%. We find that firms in the highest advertising quartile (mean annual advertising $173 million) do not see any further decreases in their liquidity risks, possibly because these firms are already well known to individual investors and have low levels of liquidity risk. Firms in the lowest quartile of advertising expenditures also do not see any benefits, because the absolute amounts of
4
advertising would still be extremely low to affect investor recognition appreciably, even after a 25% increase in the advertising budget. The main impact is felt by firms in the middle two quartiles with moderate levels of advertising expenditures, whose market value can increase by up to 1.4 % solely due to this liquidity risk effect. As expected, the cash flow impact of advertising through increased sales results in the largest change in value. However, the change in value via lower liquidity risk for firms in the middle two quartiles is about 20% of the cash flow impact on value, indicating that understanding the effect of advertising on liquidity risk and firm value is important. The results in this paper make significant contributions to the increasingly important literature studying the impact of marketing actions on firm value. First, we introduce liquidity risk as a hitherto unexplored (in the marketing literature) but important driver of firm value and suggest that marketing efforts such as product advertising can lower liquidity risk by affecting investor perception and behavior. Second, we also show that liquidity risk is negatively related to the number of individual investors. Joshi and Hanssens (2010) find that advertising has a direct effect on firm value through the spillover and signaling effects on investors. Their discussion implies that the driving force for these effects is the product market effect of advertising in making the stock more attractive to investors. Our research complements and extends the extant literature by identifying a capital market effect of advertising. The impact of marketing activities on liquidity risk has, to the best of our knowledge, not been examined in the marketing or the finance literature. Third, our results also provide greater insight into the economic significance of liquidity risk by simulating the impact of increased advertising on firm value, both through increased cash flow and reduced liquidity risk (which reduces the cost of equity capital). Finally, at a broader level, our finding that advertising lowers liquidity risk and leads to a lower cost of
5
capital may enable a firm to convert hitherto unprofitable projects (e.g., a brand extension decision, new product introduction) into viable ones. Consequently, our results indicate that the returns to advertising have likely been underestimated in the literature. The rest of the paper is organized as follows. We first provide the conceptual background and discuss how marketing actions such as advertising can lower liquidity risk. We then describe the data and the methodology used to estimate liquidity risk and discuss the main results. Next, we estimate the incremental value created by advertising via its impact on liquidity risk and conclude by discussing the managerial implications of our findings.
Advertising and Liquidity Risk
The most widely accepted theoretical model for asset pricing in the finance literature is the Capital Asset Pricing Model (CAPM), where all investors are assumed to hold a combination of a riskless asset (such as a treasury bill) and a diversified market portfolio. The allocation between these two assets reflects the investor's choice of how much risk to take. In the CAPM, the risk of a stock, popularly known as the ?market beta?, is the standardized covariance between the stock return and the market risk factor. Firm-specific risks are diversified away by the investor. However, the CAPM assumes a perfect world in which there are no transactions costs; thus, investors can diversify costlessly. In such a perfect world, the only risk that explains crosssectional stock returns is the market beta. Recent research in finance argues that in practice, the costless trading assumption of CAPM does not hold thereby suggesting that investors are also concerned about the liquidity of their portfolio.
6
Not all securities in the US stock markets trade every day. For example, of the about 5,000 stocks available in CRSP, only 80% traded on each of the approximately 250 trading days in 2007. About 5% of the stocks did not trade on 40 days or more. The lack of trading arises due to several transactions costs that reduce liquidity. These costs include (i) out-of-pocket costs such as brokerage fees, order-processing costs, etc., and (ii) opportunity costs incurred due to investors' inability to find suitable counterparties to trade with immediately. A firm with lower transaction costs is characterized as having a higher liquidity level, which is generally described as the ease of trading a stock at the market price, at low cost, in sufficient quantities quickly (Lin, Singh, and Yu 2009, Liu 2006). Investors will be concerned about liquidity and will take the entire future stream of these transaction costs into account when valuing a stock. Studies in finance have shown that investors are compensated with higher returns for investing in lessliquid assets (e.g., Amihud and Mendelson 1986). Liquidity also varies over time, both for individual stocks as well as in the overall market. This suggests that investors holding a portfolio of assets at any point of time would be concerned both with its current and future liquidity when making their investment decisions. For example, as investment opportunities change in the marketplace, investors may need to trade and readjust their portfolios. In addition, higher future consumption needs (e.g., to buy a home or to finance a child's education) may require them to liquidate some of the assets in their portfolio. Thus, in addition to the market risk, investors are also concerned with the liquidity risk of their portfolio. Pastor and Stambaugh (2003), Acharya and Pedersen (2005), Sadka (2006), and Lee (2010), among others, document that liquidity risk is an important risk factor that affects stock returns. The preceding discussion suggests that when evaluating individual stocks in their portfolio, investors would be concerned about the systematic component of a stock's liquidity
7
risk, analogous to the notion of systematic market risk. The firm-specific component would be diversified away and would not be relevant for the investor. Liquidity risk represents the nondiversifiable systematic risk that a firm's stock may become illiquid in times of market stress. Investors are compensated with a risk premium for this liquidity risk, arising due to commonality or covariation in liquidity over time (e.g., Chordia, Roll, and Subrahmanyam 2000, especially the discussion in their Section 1.2). Hence, an investor would require a smaller risk premium for a stock with low liquidity risk that contributes to reducing the overall liquidity risk of her portfolio. This lower risk premium will decrease a firm's cost of capital and increase firm value. We emphasize that liquidity level and liquidity risk represent two distinct, albeit related, aspects of liquidity that are priced by investors. Advertising can potentially impact both the liquidity level (a positive link is documented in Grullon, Kanatas and Weston (2004)) and liquidity risk of a firm. We focus on the hitherto unexplored impact of advertising on liquidity risk and present our conceptual model in Figure 1. Insert Figure 1 about here Advertising can attract investors to a firm in two ways. First, advertising can impact customer behavior and increase sales, market share, or profits of firms (Vakratsas and Ambler 1999). This makes the firm more attractive as an investment. We label this as the product market outcome of advertising. Second, media advertising intended to create awareness among consumers in general can also have a spillover impact on the investor community, resulting in higher overall awareness about the firm and its products (Lovett and MacDonald 2005, Joshi and Hanssens 2010). Traditionally, investors have been classified into two broad groups - individual investors and institutional investors such as mutual funds, hedge funds, and pension funds. Institutional investors are typically already aware of all investment opportunities, and thus they
8
may be attracted to higher advertising firms mainly due to the implied positive product market outcomes, and not due to the spillover effect of advertising. On the other hand, individual investors face time and capital constraints, and are much more likely to be affected by both increased awareness and product market signals of advertising. For example, Frieder and Subrahmanyam (2005) conclude that individual investors prefer to invest in stocks with easily recognized brands and Fehle, Tsyplakov and Zdorovtsov (2005) document a significant increase in buying activity by small investors immediately after Super Bowl advertisements. Thus, advertising likely creates awareness of and preference for the firm and helps increase the consideration set (Mitra and Lynch 1995) in which individual investors evaluate investment opportunities. Research in finance suggests that trading by individual and institutional investors affect liquidity risk differently. Trades occur when buyers and sellers exchange assets at the ongoing market price, which makes the stock liquid.ii While active institutional investors exhibit herd behavior and tend to use up liquidity by typically trading in the same direction (Dennis and Strickland 2002), individual investors step in to provide liquidity (Kaniel, Saar and Titman 2008). Thus we conjecture that a firm that has a larger pool of individual investors whose trading provides liquidity on a consistent basis will have lower liquidity risk. Investors in such stocks can be more confident that the presence of a large pool of liquidity providers will enable them to complete their trades when needed. Simply put, investors will require a smaller risk premium as compensation for having such stocks in their portfolio. If advertising is to reduce a stock's liquidity risk, the effect has to be to attract liquidity providers to the stock and ensure that the stock's liquidity does not dry up, especially in times of market stress. This can be achieved if advertising is able to attract a large number of individual
9
investors whose ability to participate in the stock market does not fluctuate greatly in good times and bad times (as discussed earlier), thus providing an ex-ante expectation of a steady source of liquidity, and lowering liquidity risk.iii This suggests that firms that advertise will have a lower liquidity risk than firms that do not. Furthermore, the level of advertising will be related to the level of liquidity risk. All else equal, firms that advertise more heavily will be able to create awareness among a larger pool of individual investors and will reduce liquidity risk to a greater extent than firms that have limited advertising budgets. In other words, as the magnitude of advertising expenditures increases, the individual investor pool will be larger, leading to lower liquidity risk. We test the following hypotheses in this paper: H1a: Firms with zero advertising expenditure will have higher liquidity risk than firms that advertise. H1b: Firms that spend more (less) in total advertising dollars will have lower (higher) liquidity risk. H2: Liquidity risk will be lower for firms that have a larger number of individual investors.
Data and Methodology
Sample Selection We use data for US based firms available in the merged CRSP COMPUSTAT file from Wharton Research Data Services (WRDS) during the 35 years from 1971 through 2005. We limit our analysis to this period as only a small number of firms reported advertising expenditure
10
before 1971 and the liquidity risk factor LIQ (Liu 2006), obtained from the author, is only available to us till 2005. We exclude foreign firms that are traded as American Depository Receipts (ADRs), closed end funds and other securities, retaining only common stocks of domestic US firms with a share code of 10 or 11. On average, about 62% of the firms have missing advertising data per year. Because we cannot reliably determine whether these firms actually had significant advertising expenditure but failed to report it separately or they did not spend any material amount on advertising, we exclude these firms from the sample. Finally, we retain only those firms in our sample that have monthly stock returns data available on CRSP. This results in an average sample size of 1,812 firms per year (63,429 firm years) representing about 30% of the COMPUSTAT firms per year. Our model estimation requires monthly data on stock returns, returns on U.S Treasury bills (the risk free rate), data on the three factors proposed by Fama and French (1993) and the momentum factor proposed by Carhart (1997); these data are all obtained from WRDS. All financial statement data are from COMPUSTAT, stock returns data are from CRSP, and ownership data are from the Thomson Reuters database. In comparison to other studies in marketing using financial data, our sample is larger in size, spans a longer time period of thirty-five years, and is broader in scope including both large and small firms. For example, McAlister, Srinivasan, and Kim (2007) analyze 644 firms over a period of 22 years (1979-2001). They limit their sample to the firms listed on New York Stock Exchange (NYSE). In contrast, our sample includes firms listed on any of the three stock exchanges - NYSE, AMEX, and NASDAQ.
11
Methodology to Estimate Liquidity Risk We use a procedure similar to the estimation of systematic risks in the Fama French model (Fama and French 1993). Typically, a risk factor is created as the difference in returns between stocks with high and low values of the variable of interest (e.g., firm size, book to market ratio, etc.). The risk level is then estimated as the covariance of a single stock's returns with the risk factor. We follow this approach and use the liquidity risk factor LIQ (Liu 2006). Liu (2006) uses his LM12 liquidity level measure to sort all firms in the market into three liquidity groups. LIQ is then estimated as the difference in returns between portfolios of stocks with low and high liquidity. Details of the measures are provided in the Web Appendix. The liquidity risk is estimated as the slope coefficient of the regression of a stock's monthly excess returns on the liquidity risk factor LIQ (and other risk factors as discussed later). Others (e.g., Eckbo and Norli 2005) use a similar approach to measure liquidity risk, except that they use turnover (number of shares traded to number of shares outstanding) as a measure of stock's liquidity level. We prefer Liu's (2006) liquidity level measure LM12 because it additionally captures the difficulty of trading (LM12 uses both the number of zero trading days and turnover); thus the LIQ measure is more comprehensive.iv In order to estimate the link between advertising and liquidity risk, we use the calendartime portfolio method (Srinivasan and Hanssens 2009). This method is commonly utilized in the finance literature and has also been used in prior studies in marketing (e.g., Sorescu, Shankar and Kushwaha 2007). Using a portfolio of firms rather than individual firms increases the precision of the estimated coefficients (Beaver, Kettler and Scholes 1970). We first rank all the stocks based on their advertising expenditures during the portfolio formation period of one year. We then either divide all the stocks into two portfolios (zero-advertising and positive-advertising, to
12
test H1a), or divide all stocks with positive advertising expenses into four equal sized portfolios (to test H1b). We reassign firms to portfolios at the end of every portfolio formation period and repeat this process over 1971-2005. We invest equal amounts in all stocks in a given portfolio during the model estimation period of one year and calculate the monthly return on the portfolio. We thus generate a time-series of monthly portfolio returns for every portfolio. In calculating the portfolio return, we also account for firms that delist from the stock exchange over time. A delisted firm could have gone bankrupt, or moved to a different exchange. If a delisted firm's return is not properly accounted for in estimating the portfolio returns, it introduces a bias in the returns. The details of this adjustment are provided in the Web Appendix. We illustrate this calendar-time portfolio procedure for firms with positive advertising expenses. We start by sorting the firms into quartiles, P1 to P4, based on the advertising expenditures in 1971. P1 (P4) consists of all the firms in the lowest (highest) advertising quartile. The corresponding model estimation period is 1972. The equal-weighted portfolio returns are calculated for every month in 1972 using the monthly returns for individual stocks in the portfolio. This procedure yields 12 monthly returns for each of the P1 to P4 portfolios, which are used to estimate the regression models (described below) for each portfolio. We repeat this procedure for the next portfolio formation period, 1972 andreassign firms to the four P1 to P4 portfolios, based on advertising expenses in 1972. Next, we calculate the monthly portfolio returns and estimate the regression model using the 12 monthly portfolio returns in the model estimation period 1973. We continue this procedure until 2005 and end up with 34 nonoverlapping model estimation periods, i.e., 1972 to 2005. As shown in Table 2, each portfolio, on average, consists of 426 firms per year. Figure 2 shows the portfolio formation and model
13
estimation periods discussed above as well as the ones used in the robustness checks discussed later. Insert Figure 2 about here In the first model (liquidity risk-augmented CAPM), we include the market risk factor and the liquidity risk factor, LIQ, as factors explaining expected return (Equation 1). We include these two risk factors because they have strong theoretical underpinnings (e.g., Sharpe 1964, Lintner 1965 for the market risk factor; Pastor and Stambaugh 2003, Acharya and Pedersen 2005, Korajczyk and Sadka 2008 for liquidity risk factor). However, research in finance and marketing (e.g. Srinivasan and Hanssens 2009) also uses four factors - market, size, book-tomarket, and momentum risk factors (Fama and French 1993; Carhart 1997) to capture expected return, even though Fama and French (2004) argue that the last three risk factors are ?brute force? constructs, with little theoretical justification. Hence, we estimate a second model (2) that includes these additional risk factors, as a robustness check. (1) Rpt - Rft = ?p + ?mpMKTRFt + ?lpLIQt + ?pt (2) Rpt - Rft = ?p + ?mpMKTRFt + ?lpLIQt + sp SMBt + hpHMLt + upUMDt + ?pt where Rpt is the return on portfolio p during month t and Rft is the return on a U.S. Treasury bill during the same period; MKTRFt is the difference in return between the risky market portfolio Rmt and the risk free return Rmt in month t, LIQt is return difference between the stocks in the low and high liquidity portfolios in month t, SMBt is the size risk factor in month t and is calculated as the difference between the returns of a portfolio of small vs. large firms, HMLt is the difference in returns of a portfolio of high and low book-to-market stocks, UMDt is the difference in returns between a portfolio of winner stocks with high prior returns and loser stocks
14
with low prior returns, , and ?pt is an error term. The main variable of interest is the liquidity risk, the estimated slope coefficient ?lp. We annually estimate both models for each estimation period from 1972 to 2005 and report the mean of these parameter estimates. Our regressions follow a procedure similar to that suggested by Fama and MacBeth (1973). The model estimation periods are always nonoverlapping periods, to minimize correlation in the estimated betas. Thus, we expect that the parameters estimated in one time period are independent of the parameters estimated in other time periods. In addition, if markets efficiently incorporate the available information into stock prices, monthly stock returns would be serially uncorrelated. Insert Table 1 about here Summary Statistics The first two panels in Table 1 report the summary statistics for firms that advertise (Panel A) and firms that do not (Panel B). We first compute the averages of all the financial variables across all the firms for each year and each portfolio, resulting in 35 mean values for each variable, and report the descriptive statistics of these 35 values. All dollar values are adjusted for inflation and reported in 2005 dollars. On average 1,706 firms reported positive advertising expenses each year. The mean (median) market value of equity is $1,592 ($1,071) million. The mean (median) advertising expenditure is substantial at $45 ($42) million. These firms, on average, spent 7% of their sales on advertising every year and have a mean age of 11.84 years. The sample in Panel B, with an average of 107 firms per year that reported zero advertising, is considerably smaller than that in Panel A. We find that these firms are smaller with a mean (median) market value of equity of $295 ($139) million. Although the mean total assets are $1,120 million, the median is just $211 million and the mean age is 6.75 years.
15
The next four panels in Table 1 report similar statistics for four quartile portfolios formed by sorting firms on their annual advertising expenditures. Each portfolio consists of an average of 426 firms every year. Portfolio P1 includes firms with the lowest amount of advertising expenses. The mean (median) annual advertising expenditure is only $.25 ($.21) million. The mean (median) annual advertising expenditure is $1.38 ($1.17) million for Portfolio P2, the second quartile, and is $6.88 ($6.47) million for portfolio P3. Finally, for the highest advertising quartile, portfolio P4, the mean (median) annual advertising expenditure is $172.99 ($163.29) million. For brevity, in the rest of this section we discuss summary statistics only for portfolio P1 (lowest advertising) and portfolio P4 (highest advertising). The firms in P1 are significantly smaller than the firms in P4. The average market value of equity of P1 is $70 million and is about 1.2% of the mean market value of equity of the firms in portfolio P4. We find a similar pattern in total assets, sales, and firm age, indicating that the firms in P1 are much smaller and younger than the firms in P4. In spite of the size differences, the average advertising-to-sales ratio for both the portfolios is about 7%, suggesting that firms in both portfolios have equal advertising intensity. However, given the small absolute level of annual advertising expenditures in P1 ($.25 million), any visibility generated due to advertising will likely be very small. On the other hand, with the same advertising intensity, the firms in P4 are likely to have a very high visibility due to their much higher level of average advertising expenditures ($172.99 million).
Results
In this section, we begin the empirical analysis by presenting the results of regressions
16
using the liquidity risk-augmented CAPM model (Model 1) and the liquidity risk-augmented Carhart 4-factor model (Model 2). We mainly discuss the results relating to liquidity risk (coefficient of the liquidity risk factor) because the other risk-factors are used as controls for other systematic risks. We also report the results of several robustness checks that we conducted to assess the reliability of our results. Insert Table 2 and Figure 3 about here
Comparison of Zero and Positive-Advertising Portfolios (Hypothesis H1a) In Table 2, we report the estimates of the two models for zero- and positive-advertising portfolios using a one-year portfolio formation and one-year estimation period. The stocks are sorted into two portfolios every year based on the advertising expense (zero and positive advertising) and the regression models are estimated using returns over the subsequent 12 months. There are 34 such model-estimation periods and Table 2 shows the mean and t-values of the betas across these 34 regressions. Assuming that the betas are stationary and serially uncorrelated, we can use a standard t-test to check the significance of the betas, and a paired ttest to test for differences of the betas between portfolios (Litzenberger and Ramaswamy 1979). In both models, liquidity risk is statistically significant both for firms with zero advertising and for firms that report positive advertising expenses. However, consistent with hypothesis H1a, we find that firms that do not advertise are subject to significantly higher liquidity risk than firms that use advertising. Specifically, using Model 1, liquidity risk for firms in the positiveadvertising portfolio averages .43, which is significantly lower than the mean of .91 for firms in the zero-advertising portfolio (p< .01). The results using Model 2 are similar. The corresponding estimates of liquidity risk are .38 and .96, and the mean difference is statistically significant (p<
17
.05). Figure 3 shows the difference in liquidity risk (beta for the zero-advertising portfolio - beta for the positive-advertising portfolio) for Model 1 for each year. The difference is positive in most of the years, as expected. These results indicate that the firms that do not advertise are subject to higher liquidity risk, and support hypothesis H1a. We conduct three additional tests to assess the robustness of our results. First, we extend the portfolio formation period from one to three years. We assign firms to portfolios based on the average advertising expense over the prior three years rather than just in the prior year. This is more likely to reflect the firm's longer term advertising strategy. Further, when we use a oneyear portfolio formation period, we are assuming that most of the effect of advertising on investors occurs during the next year. However if advertising has a longer term effect, then the impact on investors in period t+1 could be due to advertising expenditures in periods t, t-1, t-2 etc. The three-year portfolio formation period accounts for this longer term effect of advertising. We estimate the models in the 32 one-year estimation periods. The results are similar to those reported earlier (difference in liquidity risk; Model 1, .472, p< .05; Model 2, .649, p=.01). Second, we control for firm size because large firms tend to spend more on advertising as compared to small firms. One possible objection to forming portfolios based on advertising expenditure is that it could be equivalent to sorting them on firm size. In addition, larger firms may be more visible to investors as they are more likely to be covered by financial analysts and media. Although we control for size effects in model 2 by including the size factor SMB, we conduct an additional test to ensure that the differences in liquidity risks are primarily due to differences in advertising expenditure. For each firm in the zero-advertising sample, we select a matching firm of similar size from the positive-advertising sample that has total assets within 70130% of the total assets of zero-advertising firm (similar cutoffs are used in Barber and Lyon
18
(1997)). Using a one-year portfolio formation period and one-year estimation period, we continue to find that the liquidity risk for the zero-advertising portfolio is higher than that for the positive-advertising portfolio (difference in liquidity risk; Model 1, .276, p= .077; Model 2, .449, p<.05). We obtain qualitatively similar results where we use a three-year-portfolio formation one-year estimation window (difference in liquidity risk; Model 1, .265, p=.246; Model 2, .924, p= 0.05). This is a stronger test of hypothesis H1a, suggesting that even after explicitly controlling for firm size, advertising reduces liquidity risk. Finally, we pool all the monthly observations to obtain more precise estimates of liquidity risk, and test for differences between the liquidity risk of the two portfolios. For the one-year portfolio formation period, we pool 408 monthly returns and find that the differences are statistically significant (Model 1, .395, p<.01; Model 2, .309, p<.01). For the three-year portfolio formation method we pool 384 monthly returns and again find statistically significant differences (Model 1, .465, p< .01; Model 2, .362, p= .01). Using size matched firms, the pooled sample of both one year portfolio formation (Model 1, .282, p<.01; Model 2, .337, p<.01) and three-year formation period (Model 1, .239, p< .1; Model 2, .294, p< .1) shows significant differences. Overall, the results so far strongly support H1a that firms with positive advertising have lower liquidity risks. In the next section we test whether the amount of advertising is linked monotonically to liquidity risk (H1b). We expect that as firms advertise more, they will have a larger individual investor pool and thus lower liquidity risk.
Comparison of the Four Positive-Advertising Quartile Portfolios (Hypothesis H1b) In Table 3 we report the estimates of the models for the four positive-advertising portfolios (P1 to P4). The first four columns show the values of the risk coefficients for the four
19
portfolios. Two important observations about the estimates of liquidity risk emerge. First, using both estimation models, the liquidity risk of the low advertising portfolio P1 is always the highest and it decreases monotonically as we move from the lowest advertising to the highest advertising portfolio. This observation strongly supports hypothesis H1b, that higher levels of advertising are associated with lower liquidity risk. Second, the liquidity risk of P4 is marginally significant in model 1 (p< .1) and is not significantly different from zero in model 2. Firms spending large amounts on advertising are very visible to investors and will have large individual investor pools, and thus face very low or no liquidity risk. The last three columns in the table show that the liquidity risk of P1 firms is significantly higher than the liquidity risk of firms in portfolios P2, P3, and P4. For example, for model 1 the mean differences in the liquidity risk for P1-P2, P1-P3, and P1-P4 are respectively .380 (p< .01), .603 (p< .01) and .757 (p< .01), and are highly significant. In Figure 4 we plot the differences in liquidity risk (P1 - P2), (P1 - P3), and (P1 - P4) for Model 1 and confirm that the differences are positive in the overwhelming majority of the years. The consistently positive differences between liquidity risk of the lowest (P1) and higher advertising portfolios (P2, P3, and P4) indicate that our results are robust and not restricted to a few years. Insert Table 3 and Figure 4 about here We conduct additional robustness checks similar to those discussed earlier for zero and positive advertising portfolio comparisons. We use a three-year portfolio formation period, sizematched portfolios, and a pooled regression. We find that our results generally hold. The differences in liquidity risk are insignificant only in some of the size matched portfolios using one or three year portfolio formation periods. However, the majority of the differences in
20
liquidity risk (76%) are both in the correct direction and statistically significant at the 10% level or better, providing strong evidence in support of hypothesis H1b. Decile Portfolio Analysis As documented in Table 3, the liquidity risk of portfolio P4 stocks (with an average advertising expenditure of $173 million) is close to zero. In order to more closely examine the level of advertising at which liquidity risk becomes small, we repeat the analysis using one-year portfolio formation and one-year estimation windows by creating deciles rather than quartiles based on advertising expenditures. The results confirm that liquidity risk decreases monotonically with increasing advertising expenditures. We plot liquidity risk against median advertising expenditure for each of the ten portfolios (Figure 5), and find that liquidity risk drops sharply as advertising expenditures increases to about $7 million. Subsequent increases in advertising (until about $40 million) only gradually reduce liquidity risk. Increasing advertising beyond that does not lower liquidity risk by much. However, firms may advertise beyond this level with a view to increasing consumer awareness in product markets. Insert Figure 5 about here Overall, the results in this section strongly support hypotheses H1a and H1b, indicating that as the level of advertising increases, the liquidity risk decreases.
Firm Level Analysis The analysis thus far has used portfolios of stocks to document that advertising is related to liquidity risk. However, as the stocks are grouped into portfolios, we cannot use crosssectional firm level regression models to explore the relationship between advertising, number of investors and liquidity risk. In addition, firm level regressions will allow us to control for a
21
number of firm characteristics that may be driving cross-sectional differences in firm liquidity risk. We also use these regressions in performing economic impact analysis later in the paper. We adopt a two-step methodology used in earlier research (e.g., McAlister, Srinivasan and Kim 2007) to link advertising expenditure with liquidity risk for each firm from 1971 to 2005. To illustrate, for a given year say 1973, we regress the monthly excess return for each firm during the previous 36 months (1971-73) on the market and liquidity risk factors (liquidity risk augmented CAPM, eq. (1)). This generates a slope coefficient on the liquidity risk factor LIQ as an estimate of liquidity risk for the firm. In the second step, we regress the firm-specific liquidity risk on the firm's advertising; and include total assets, leverage, book-to-market, age, and year dummies, as control variables. Every year, we reassign firms to four advertising portfolios based on the advertising in the prior year (e.g., 1972) and re-estimate the regression models. All independent variables are lagged by one year. Preliminary tests on the residuals from our model indicate significant autocorrelation in the errors. Therefore, we use a fixed-effect panel model with AR(1) serial error correlation. The results of the second stage regression linking liquidity risk to advertising are shown in Table 4. Advertising is significantly related to liquidity risk for firms in three of the four portfolios P2, P3 and P4. Advertising has no impact on liquidity risk for firms in P1, the lowest advertising portfolio, where the coefficient on advertising is not statistically significant. The highest impact of advertising is in P2 (?=-.078, p <.05). The impact reduces in P3 (?=-.014, p<.05). We expected the coefficient in P4 to be zero as the investor awareness about these firms should be high and advertising should not impact liquidity risk. We find that the coefficient on advertising is positive and significant (?=.00047, p<.05), even though the economic magnitude of the coefficient is small.
22
Insert Table 4 about here Overall, the results in this section indicate that if the advertising levels are either too low or too high, advertising does not reduce liquidity risk. However in P2 and P3, as the level of advertising increases, liquidity risk decreases. This is likely to occur if higher advertising results in a larger numbers of individual investors. In the next section, we present evidence to test hypothesis H2, and examine whether the number of individual investors in a firm is negatively linked to liquidity risk.
Advertising, Ownership Structure and Liquidity Risk (Hypothesis H2) We examine whether an increased presence of individual investors in a firm is associated with lower liquidity risk as a direct test of hypothesis H2. We have discussed prior research (e.g., Kaniel, Saar, and Titman 2008) which documents that while institutional investors demand immediacy of trade and use up liquidity, individual investors step in and provide liquidity. Thus firms that have a larger number of individual investors would have lower liquidity risk (H2). As before, we follow a two step procedure to test this hypothesis. In the first step, we estimate the annual liquidity risk for each firm using the liquidity risk augmented CAPM over the previous 36 months. We then use these firm level betas (the liquidity risk estimates) as dependent variables in second stage regressions where the main explanatory variable of interest is the number of individual investors. The data on number of institutional investors are only available to us from 1997 and the tests in this section use this subset of data. The number of institutional investors in a given year is obtained as an average of the quarterly reported numbers. Thomson Reuters does not report the number of individual investors. We use the difference between the number of total shareholders
23
(from COMPUSTAT) and the number of institutional investors as an estimate of the number of individual investors for a firm-year. We regress our liquidity risk estimate on the log of the individual investors and include other control variables, viz., log of number of institutional investors, total assets, book-to- market, leverage, firm age, profit margin, market share, dummy variables for BtoB or BtoC firms, and year dummies. We have a total of 23,013 firm-year observations from 5,212 unique firms with non-missing observations on all the model variables from 1997 to 2005. All independent variables are lagged by one year. Preliminary tests on the residuals from our model indicate significant autocorrelation in the errors. Therefore, we use a fixed-effect panel model with AR(1) serial error correlation. The results are shown in Table 5. We find that the liquidity risk of firms is significantly negatively related to the number of individual investors, lending support to H2. Interestingly, we find that liquidity risk is significantly positively related to the number of institutional investors. The opposite signs on individual and institutional ownership suggest that only individual ownership is associated with lower liquidity risk, consistent with the assertion in the finance literature.v Insert Table 5 about here We conducted three further tests to examine whether the effectiveness of advertising in lowering liquidity risk occurs via creating awareness among individual investors (these results are not tabulated separately, but are available from the authors upon request). First, if advertising is effective in reducing liquidity risk for a firm due to its ability to attract individual investors, then we should expect advertising to have a greater impact in BtoC firms (where the advertising is targeted at individuals) rather than in BtoB firms (where the advertising is not targeted towards individuals). Second, investors are more likely to be aware of older firms that have been in existence for a long time, as compared to younger firms. Hence, we should expect advertising to
24
be more effective in reducing liquidity risk in the younger firms in P2 and P3. Third, if advertising reduces liquidity risk via its effect on individual investors, a mediation analysis would provide additional confirming evidence.
BtoB versus BtoC firms. Compared to a BtoC firm, if the advertising of a BtoB firm is targeted mainly towards trade journals and other media that have limited circulation among individuals, it will create less awareness among individual investors and we would expect it to be less likely to reduce liquidity risk. We create two size and advertising expenditure matched portfolios representing these two types of firms. Fama and French (1997) use four digit SIC codes to develop 48 industry categories that are used widely in finance research to form industry portfolios.vi We select firms that are in typically B2B (wholesale, machinery, lab equipment, construction material, and trading industries) or B2C (candy and soda, tobacco, consumer goods, and retail categories) industries. For each firm in the B2B category, we find a matching firm in the B2C category whose total assets and advertising expenses are within 70% to 130% of the B2B firm. We end up with an average of 104 matched firms per year in each of the B2B and B2C groups. We then divide the firms into two equal-sized groups based on their advertising expenditure. Thus, for each year between 1971 and 2004 we form two portfolios within the two industry groups. In both B2B and B2C industries, the average advertising for low- (high-) advertising portfolio firms is about $1.3 million ($25 million) and the total assets average about $80 million ($900 million). We expect that compared to B2B firms, the liquidity risk for B2C firms should decrease more as we move from the low-advertising to the high-advertising portfolio. Advertising by B2C firms should be visible to more investors as compared to advertising by B2B firms, since the
25
audience for advertising by B2C firms is typically broader. Over the 34-years estimation period (1972-2005), the difference between the high and low advertising portfolios is positive and usually statistically significant for annual (difference in liquidity risk; B2C firms, .341, p< .01; B2B firms, .357, p< .01) as well as pooled regressions (difference in liquidity risk; B2C firms, .187, p< .01; B2B firms, .028, ns). While the reduction in liquidity risk for BtoB and BtoC firms is similar when we use annual regressions, the reduction in liquidity risk is more for B2C firms than B2B firms using pooled regressions (difference in decrease in liquidity risk between B2B and B2C, .159, p< .05). These results offer some support for the notion that the visibility of the advertising to investors helps improve investor awareness and lowers liquidity risk.
Young versus old firms: We examine the impact of age by re-estimating the models in Table 4 separately for younger (age <= median age) and older (age > median age) firms in each of the four portfolios. The dependent variable is liquidity risk and the main explanatory variable of interest is advertising. We find that the coefficient on advertising is significantly negative only for the younger firms in P2 (-.154, p<.01) and in P3 (-.018, p<.05). The coefficient on advertising is negative but not significant at the usual levels for the older firms. As before, we continue to find that advertising does not lower liquidity risk for either young or old firms in P1 and P4; the coefficients on advertising in all these four models are not significantly different from zero. Thus, advertising appears to be effective in lowering liquidity risk only for younger firms, suggesting that improving awareness among individual investors and attracting them to relatively unknown firms is how advertising helps in reducing liquidity risk.
26
Mediation analysis: We conduct a mediation analysis to investigate whether the number of individual investors mediates the effect of advertising on liquidity risk. We test the conceptual model in Figure 1 using both the Baron and Kenny (1986) approach and the structural equations modeling approach (Zhao, Lynch and Chen 2010). We transform our panel data to meet the assumptions of these approaches, and confirm that the number of individual investors indeed fully mediates the link between advertising and LR. We also find that the number of institutional investors does not mediate liquidity risk. Taken together, these results strongly support H2 that liquidity risk will be lower for firms with more individual investors and that advertising likely increases the awareness of individual investors, and attracting them to invest in the firm
Economic Impact We have shown that advertising is negatively related to liquidity risk. However, increased advertising can create value in several ways, such as 1) increasing sales and thus cash flow, 2) creating brand loyalty, reducing cash flow variability and therefore decreasing systematic market risk, 3) increasing the liquidity level for a firm's stock and 4) increasing individual investor awareness and reducing liquidity risk. A reduction in market risk or liquidity risk and an increase in liquidity level will all lead to a lower cost of equity. Whether or not the incremental valuation impact of the reduction in liquidity risk is large will determine the importance of these results for practicing managers. We make some simplifying assumptions and estimate the partial impact of each effect on firm value one at a time (holding the other three effects constant). We estimate the impact of a 25% increase in advertising expenditure on market value for a typical firm in each of the four positive advertising portfolios.
27
To estimate the impact on sales, we assume that the advertising elasticity of demand is .2 (as in Leone 1995) for the average firm in our sample. Of course, we note that the effect on cash flow for an individual firm will be different and will depend on its advertising elasticity. The increased cash flows of the average firm in the portfolio are calculated as (operating margin*sales), net of advertising expenditure and taxes. We assume that the corporate tax rate is 35%. In order to estimate the impact on liquidity level we use estimates from Grullon, Kanatas and Weston (2004, average relative bid-ask spread calculated based on Table 4 panel A, advertising coefficient from Table 5 column 2). To assess the impact of increased advertising on reduction in market risk, we use the regression coefficients of market risk on advertising from McAlister, Srinivasan and Kim (2007, Table 2 column 1). The change in the firm's cost of equity capital in each portfolio due to a 25% increase in advertising is determined by multiplying the new market and liquidity risk by the average risk premium over our sample period. The firm value is estimated by dividing the cash flows by the estimated cost of equity (i.e., assuming perpetual cash flows). Insert Table 6 about here We first estimate the marginal impact of advertising on firm value via the cash flow effect, holding the cost of equity constant. In all four portfolios, advertising has a large impact on firm value through increased cash flows. This is expected since the first order impact of advertising on firm value is via increased sales. The cash flow impact on firm value varies between 4.63% (P1) and 6.62% (P3). Similarly, we estimate marginal changes in firm value due to reduction in cost of equity, assuming that there is no change in net cash flow. We note that this assumption implies that advertising increases sales to at least offset the increased cost of advertising. vii The impact on value due to reduction in market risk is much smaller and varies
28
between .09% (P1) and 1.10% (P4). Increases in liquidity levels are also associated with a small increase in firm value between .23% (P1) and .37% (P4). The increase in value due to change in liquidity risk varies between 0% (P1) to 1.39% (P3). In P1 there is no change in firm value due to reduction in liquidity risk. This result is not surprising when we consider that even a 25% increase in advertising is an economically small amount and is unlikely to increase individual investor awareness. In contrast, firms in the highest advertising quartile are already likely to be quite visible to the investors. Any increase in such firms' advertising expenditure is unlikely to increased investor awareness. Hence, for these firms also there should not be any marginal reduction in liquidity risk due to increased advertising. However, as discussed earlier, the small increase in liquidity risk leads to an estimated reduction of 1.27% in value. We note that these firms may still be creating incremental value by increasing advertising because the impact on value via increased cash flows is larger in magnitude (5.29%). The more interesting cases are the firms in P2 and P3, with medium levels of advertising. The increase in firm value due to reduction in liquidity risk is 1.30% in P2 and 1.39% in P3. This indicates that out of the four advertising portfolios, the firms in portfolio P3 gain the most due to the reduction in liquidity risk followed by the firms in portfolio P2. Based on P2's original valuation of $165.32 million an increase of 1.30% corresponds to an increase in market value of around $2.15 million. Similarly, from the base value of $455.91 million, the increase in P3's market value is around $6.34 million. Considering P2 and P3's incremental advertising expenditure of $ .35 million and $1.72 million respectively, the increase in value due to reduced liquidity risk is substantial. This suggests that the value relevance of advertising may have been underestimated in prior research.
29
Discussion and Conclusions
In this paper, we hypothesize and document that advertising can reduce liquidity risk through its impact on individual investor behavior. While Grullon, Kanastas and Weston (2004) examine the impact of advertising on liquidity level, our paper is the first to document its impact on liquidity risk. An economic impact analysis shows that while the cash flow impact of increasing advertising is large, reduction in liquidity risk also improves firm value. For a typical firm in quartiles P2 and P3, the improvement in firm value due to lower liquidity risk is about four to five times that due to increase in the liquidity level. Thus, it is important for marketers to understand not only the impact of advertising on cash flow, but also on liquidity risk. Our main contributions to the marketing literature are (a) to introduce liquidity risk as an important driver of firm value that can be affected by marketing actions such as advertising, (b) to show that the spillover effect of advertising (Lovett and MacDonald 2005, Joshi and Hanssens 2010) plays an important role in affecting individual investor awareness in capital markets, and (c) to suggest that methodologically, future researchers should include liquidity risk (which is a priced, systematic risk) in their models of expected returns. This reinforces the call of Tuli and Bharadwaj (2009) for empirical literature in marketing to focus on multiple dimensions of risk including systematic, idiosyncratic and downside risk. Managerial implications. Our results suggest that firms can increase value by attracting and retaining individual investors. Firms typically communicate with the investor community through the required disclosures of financial information (e.g., financial statements, 10K reports, etc.) as well as by establishing an investor relations (IR) strategy with the ?goal of creating an understanding of the firm, attracting information intermediaries, and targeting a desired investor
30
base? (Brennan and Tamarowski 2000). The creation of the IR department results in increased firm visibility through higher media coverage and analyst following as well as more institutional investors (Bushee and Miller 2007). While the IR activities mainly focus on the institutional investor community, our findings suggest that IR departments should also pay attention to communicating with individual investors. Marketing managers make decisions both on the level of advertising expenditures as well as on the media-mix that determines the allocation of this money across various media. It appears that advertising is effective in reducing liquidity risk for firms (in P2 and P3) that have annual advertising budget between about $ 0.5 million and $20 million. Firms spending outside this range will not see much reduction in their liquidity risk due to increased ad-spending. This impact may be especially important for younger firms, as well as BtoC firms as their advertising can be more effective in reducing liquidity risk. Even if the advertising budgets cannot be increased, our results imply that firms can carefully select the advertising media mix to achieve their goal of targeting individual investors. Specifically, if firms can redirect some of the product advertising to media that simultaneously targets individuals who are potential investors, this could also increase awareness among individual investors. Firms may advertise their products directly in the financial press (e.g., Wall Street Journal) or in media that are commonly perused by individual investors (e.g., CNBC, local print media, etc.). For example, Kodak advertised their ink saving inkjet printers and Sony advertised its e-book reader using full page advertisements in the Wall Street Journal (March 30, 2009). Firms may also be able to develop specific advertising and marketing campaigns aimed at individual investors and encourage them to invest in the firm. This may be even more important in times of market stress. In addition to media mix, the timing of the advertising may also impact liquidity risk. Media planners use
31
pulsing of advertising expenditures to increase their effectiveness (Leckenby and Kim 1994). But this can result in cash flow volatility, leading to higher risk (Fischer, Shin and Hanssens 2009). Analogously, pulsing might also increase the fluctuations in awareness among individual investors, which could lead to increased liquidity risk. Managers may need to take this additional negative financial impact into consideration as they evaluate the timing of their marketing spending policies. Our results also highlight a crucial difference between liquidity level and liquidity risk. Specifically, if a firm is only interested in increasing the liquidity level of its stock, then ownership structure may be less important and it may not matter whether the higher liquidity level comes about due to increased trading by institutional or individual investors. However our results indicate that a larger pool of institutional (individual) investors increases (decreases) liquidity risk, which may reduce (augment) any benefit the firm may obtain by increasing liquidity level. Since the impact of reducing liquidity risk on value is substantial compared to that of increasing liquidity level, our results suggest that firms should focus more on increasing the number of individual investors. This would both increase the liquidity level and decrease liquidity risk. Our findings, coupled with other findings about impact of advertising on systematic risk (McAlister, Srinivasan and Kim 2007), should also make it easier for marketing executives to justify their advertising expenditures to the financial executives, who are typically skeptical about the financial returns to these investments. We believe that our findings about the relation between liquidity risk and investor type are also new to the finance literature. By documenting that advertising can help lower the cost of equity capital by lowering both market and liquidity risks, the marketing department can more accurately quantify the full value of advertising and
32
related marketing expenditures. Managers may be able to reduce the cost of equity by keeping the advertising budget the same but altering the media mix. The lower cost of capital would improve the viability of hitherto unprofitable marketing projects such as brand extensions, new product introductions, etc. Limitations and future research. The effectiveness of the firm in raising individual investor awareness through product market advertising will depend on the branding strategy of the firm. Advertising should be more effective for firms that follow a corporate branding strategy compared to a house-of-brands strategy, as investors can link the brands and products of the corporate brand with the firm more easily. It was not possible for us to code the branding strategy for such a large number of firms, and we leave this to future research. The efficacy of different types of media, e.g., TV, print, internet, etc., is also likely to be different in raising investor awareness. A number of other marketing strategies, besides advertising, can also impact investor awareness. For example, a firm targeting a broader market segment or with an intensive distribution is likely to be more visible to investors. Similarly, the CSR actions of a firm can increase the likelihood that socially responsible individual investors would be willing to invest in the firm. The capital market impact of these and other marketing actions need to be studied more carefully to fully understand the impact of marketing actions on liquidity risk and firm value.
33
References Acharya, Viral V. and Lasse Heje Pedersen (2005), ?Asset Pricing with Liquidity Risk,? Journal of Financial Economics, 77 (2), 375-410.
Amihud, Yakov (2002), ?Illiquidity and Stock Returns: Cross-Section and Time-Series Effects,? Journal of Financial Markets, 5 (1), 31-56.
—— and Haim Mendelson (1986), ?Asset Pricing and the Bid-Ask Spread,? Journal of Financial Economics, 17 (2), 223-49.
Barber, Brad M. and John D. Lyon (1997), ?Detecting Long-Run Abnormal Stock Returns: The Empirical Power and Specification of Test Statistics,? Journal of Financial Economics, 43 (3), 341-72.
Baron, R. M. and D. A. Kenny (1986), ?The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations,? Journal of Personality and Social Psychology, 51 (6), 1173-82.
Beaver, William, Paul Kettler, and Scholes Myron (1970), ?The Association between Market Determined and Accounting Determined Risk Measures,? The Accounting Review, 45 (4), 65482.
34
——, Maureen McNichols, and Richard Price (2007), ?Delisting Returns and Their Effect on Accounting-Based Market Anomalies,? Journal of Accounting and Economics, 43 (2-3), 341-68. Brennan, Michael J. and Claudia Tamarowski (2000), ?Investor Relations, Liquidity, and Stock Prices,? Journal of Applied Corporate Finance, 12 (4), 26-37.
Bushee, Brian J. and Gregory S. Miller (2007), ?Investor Relations, Firm Visibility, and Investor Following,? SSRN eLibrary Working Paper.
Carhart, Mark M. (1997), ?On Persistence in Mutual Fund Performance,? Journal of Finance, 52 (1), 57-82.
Chordia, Tarun, Richard Roll, and Avanidhar Subrahmanyam (2000), ?Commonality in Liquidity,? Journal of Financial Economics, 56 (1), 3-28.
Dennis, Patrick J. and Strickland Deon (2002), ?Who Blinks in Volatile Markets, Individuals or Institutions?,? The Journal of Finance, 57 (5), 1923-49.
Eckbo, B. Espen and Øyvind Norli (2005), ?Liquidity Risk, Leverage and Long-Run Ipo Returns,? Journal of Corporate Finance, 11 (1-2), 1-35.
Fama, Eugene F and Kenneth R. French (1993), ?Common Risk Factors in the Returns on Stocks and Bonds,? Journal of Financial Economics, 33 (1), 3-56.
35
—— and —— (1997), ?Industry Costs of Equity,? Journal of Financial Economics, 43 (2), 15393.
—— and —— (2004), ?The Capital Asset Pricing Model: Theory and Evidence,? Journal of Economic Perspectives, 18 (3), 25-46.
—— and James D. MacBeth (1973), ?Risk, Return, and Equilibrium: Empirical Tests,? Journal of Political Economy, 81 (3), 607-36.
Fehle, F., S. Tsyplakov, and V. Zdorovtsov (2005), ?Can Companies Influence Investor Behaviour through Advertising? Super Bowl Commercials and Stock Returns,? European Financial Management, 11 (5), 625-47.
Fischer, Marc, Hyun Shin, and Dominique M. Hanssens (2009), ?Marketing Spending and the Volatility of Revenues and Cash Flows,? Working Paper (UCLA Anderson School of Management).
Fornell, Claes, Sunil Mithas, Forrest V. Morgeson III, and M. S. Krishnan (2006), ?Customer Satisfaction and Stock Prices: High Returns, Low Risk,? Journal of Marketing, 70 (1), 3-14.
Frieder, Laura and Avanidhar Subrahmanyam (2005), ?Brand Perceptions and the Market for Common Stock,? Journal of Financial and Quantitative Analysis, 40(1), 57-85.
36
Gruca, Thomas S. and Lopo L. Rego (2005), ?Customer Satisfaction, Cash Flow, and Shareholder Value,? Journal of Marketing, 69 (3), 115-30.
Grullon, Gustavo, George Kanatas, and James P. Weston (2004), ?Advertising, Breadth of Ownership, and Liquidity,? Review of Financial Studies, 17 (2), 439-61.
Hanka, Gordon (1998), ?Debt and the Terms of Employment,? Journal of Financial Economics, 48 (3), 245-82.
Huberman, Gur (2001), ?Familiarity Breeds Investment,? Review of Financial Studies, 14 (3), 659-80.
Joshi, Amit and Dominique M. Hanssens (2009), ?The Direct and Indirect Effects of Advertising Spending on Firm Value,? Journal of Marketing, 74 (1), 20-33.
Kaniel, Ron, Gideon Saar, and Sheridan Titman (2008), ?Individual Investor Trading and Stock Returns,? The Journal of Finance, 63 (1), 273-310.
Korajczyk, Robert A. and Ronnie Sadka (2008), ?Pricing the Commonality across Alternative Measures of Liquidity,? Journal of Financial Economics, 87 (1), 45-72.
Leckenby, John D. and Heejin Kim (1994), ?How Media Directors View Reach/Frequency Estimation: Now and a Decade Ago,? Journal of Advertising Research, 34 (5), 9-21.
37
Lee, Kuan-Hui (2010), ?The World Price of Liquidity Risk,? Journal of Financial Economics, Forthcoming.
Leone, Robert P. (1995), ?Generalizing What Is Known About Temporal Aggregation and Advertising Carryover,? Marketing Science, 14 (3), G141-G50.
Lin, Ji-Chai, Ajai K. Singh, and Wen Yu (2009), ?Stock Splits and the Trading Speed Improvement Hypothesis,? Journal of Financial Economics, 93 (3), 474-89.
Lintner, John (1965), ?The Valuation of Risk Assets and the Selection of Risky Investments in Stock Portfolios and Capital Budgets,? The Review of Economics and Statistics, 47 (1), 13-37.
Litzenberger, Robert H. and Krishna Ramaswamy (1979), ?The Effects of Dividends on Common Stock Prices: Theory and Empirical Evidence,? Journal of Financial Economics, 7 (2), 163-95.
Liu, Weimin (2006), ?A Liquidity-Augmented Capital Asset Pricing Model,? Journal of Financial Economics, 82 (3), 631-71.
Lovett, Mitchell J. and Jason B. MacDonald (2005), ?How Does Financial Performance Affect Marketing? Studying the Marketing-Finance Relationship from a Dynamic Perspective,? Journal of the Academy of Marketing Science, 33 (4), 476-85.
38
Lundgren, Terry J. (2007), ?Federated Plans Corporate Name Change,? (accessed April 22, 2009), [available at http://phx.corporate-ir.net/phoenix.zhtml?c=84477&p=irolnewsArticle&ID=967632&highlight=].
Luo, Xueming (2007), ?Consumer Negative Voice and Firm-Idiosyncratic Stock Returns,? Journal of Marketing, 71 (3), 75-88.
——and CB Bhattacharya (2009), ?The Debate over Doing Good: Corporate Social Performance, Strategic Marketing Levers, and Firm-Idiosyncratic Risk,? Journal of Marketing, 73 (6), 198-213.
McAlister, Leigh, Raji Srinivasan, and Min Chung Kim (2007), ?Advertising, Research and Development, and Systematic Risk of the Firm,? Journal of Marketing, 71 (1), 35-48.
Merton, Robert C. (1987), ?A Simple Model of Capital Market Equilibrium with Incomplete Information,? Journal of Finance, 42 (3), 483-510.
Mitra, Anusree and John G. Lynch, Jr. (1995), ?Toward a Reconciliation of Market Power and Information Theories of Advertising Effects on Price Elasticity,? Journal of Consumer Research, 21 (4), 644-59.
39
Mizik, Natalie and Robert Jacobson (2007), ?Myopic Marketing Management: Evidence of the Phenomenon and Its Long-Term Performance Consequences in the Seo Context,? Marketing Science, 26 (3), 361-79.
Pástor, Luboš and Robert F. Stambaugh (2003), ?Liquidity Risk and Expected Stock Returns,? Journal of Political economy, 111 (3), 642-85.
Sadka, Ronnie (2006), ?Momentum and Post-Earnings-Announcement Drift Anomalies: The Role of Liquidity Risk,? Journal of Financial Economics, 80 (2), 309-49.
Schwartz, Jonathan (2007), ?The Rise of JAVA - the Retirement of SUNW,? (accessed April 22, 2009), [available at http://blogs.sun.com/jonathan/entry/java_is_everywhere].
Sharpe, William F. (1964), ?Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk,? The Journal of Finance, 19 (3), 425-42.
Shumway, Tyler (1997), ?The Delisting Bias in Crsp Data,? Journal of Finance, 52 (1), 327-40.
—— and Vincent A. Warther (1999), ?The Delisting Bias in Crsp's Nasdaq Data and Its Implications for the Size Effect,? The Journal of Finance, 54 (6), 2361-79.
40
Sorescu, Alina, Venkatesh Shankar, and Tarun Kushwaha (2007), ?New Product Preannouncements and Shareholder Value: Don't Make Promises You Can't Keep,? Journal of Marketing Research, 44 (3), 468-89.
—— and Jelena Spanjol (2008), ?Innovation's Effect on Firm Value and Risk: Insights from Consumer Packaged Goods,? Journal of Marketing, 72 (2), 114-32.
Srinivasan, Shuba and Dominique M. Hanssens (2009), ?Marketing and Firm Value: Metrics, Methods, Findings, and Future Directions,? Journal of Marketing Research, 46 (3), 293-312.
Tuli, Kapil R. and Sundar G. Bharadwaj (2009), ?Customer Satisfaction and Stock Returns Risk,? Journal of Marketing, 73 (6), 184-97.
——, Sundar G. Bharadwaj, and Ajay K. Kohli (2010), ?Ties That Bind: The Impact of Multiple Types of Ties with a Customer on Sales Growth and Sales Volatility,? Journal of Marketing Research, 47 (1), 36-50.
Vakratsas, Demetrios and Tim Ambler (1999), ?How Advertising Works: What Do We Really Know?,? Journal of Marketing, 63 (1), 26-43.
Zhao, Xinshu, John G. Lynch, Jr., and Qimei Chen ?Reconsidering Baron and Kenny: Myths and Truths About Mediation Analysis,? Journal of Consumer Research, 37 (August), 197-206.
41
TABLE 1 Descriptive Statistics of the Sample Firms (1971 through 2005) MVE (Mill. $) Total Assets (Mill. $) Sales Advertising Advertising /Sales R&D /Sales Leverage Firm Age (Years) 11 .84 10 .89 11 .79 Q3 13 .08 N 1705.69 6.75 5.72 6.26Q3 7.52N 106.57 6.11 4.88 6.14Q3 7.02N 426.03 7.84 7.13 7.74Q3 8.38N 426.49 10 .73 9.48 10 .72 12 .09 N 426.89 22 .67 21 .30 22 .12 24 .86 N 426.29
Mean Q1 Median
Mean Q1 Median
Mean Q1 Median
Mean Q1 Median
Mean Q1 Median Q3
Mean Q1 Median Q3
(Mill. $) (Mill. $) A: Firms with Positive Advertising 1592.12 2253.57 1551.89 45 .36 0.07 0.58 710.80 1069.43 1306.83 36 .75 0.03 0.05 1070.94 1699.07 1425.97 42 .32 0.04 0.23 2308.84 2298.88 1806.41 57 .97 0.07 0.57 1705.69 1705.69 1705.69 1705.69 1705.69 740.80 B : Firms with Zero Advertising 295.14 1120.18 126.49 0.00 0.00 2.48 67 .09 92 .33 41 .39 0.00 0.00 0.05 139.31 211.37 85 .57 0.00 0.00 0.61 331.14 1674.37 228.70 0.00 0.00 3.26 106.57 106.57 106.57 106.57 106.57 7.54 C: Firms in the First Advertising Quartile (Lowest Advertising) - Portfolio P1 70 .15 115.98 61 .70 0.25 0.07 1.45 34 .27 26 .09 19 .89 0.15 0.01 0.11 51 .94 129.70 38 .93 0.21 0.04 0.53 77 .17 178.92 87 .73 0.28 0.06 1.47 426.03 426.03 426.03 426.03 426.03 206.86 D: Firms in the Second Advertising Quartile - Portfolio P2 165.32 295.44 148.48 1.38 0.09 0.59 75 .59 96 .30 85 .37 0.92 0.02 0.03 108.87 240.68 114.41 1.17 0.04 0.16 240.68 511.34 188.85 1.60 0.07 0.39 426.49 426.49 426.49 426.49 426.49 198.37 E: Firms in the Third Advertising Quartile - Portfolio P3 455.91 773.92 464.94 6.88 0.05 0.13 235.79 388.73 272.66 5.23 0.03 0.02 286.79 528.35 436.77 6.47 0.04 0.07 641.03 960.34 614.96 7.45 0.05 0.16 426.89 426.89 426.89 426.89 426.89 163.03 F: Firms in the Fourth Advertising Quartile (Highest Advertising) - Portfolio P4 5679.30 7831.36 5534.03 172.99 0.07 0.05 2410.23 3621.39 4574.24 137.71 0.05 0.03 3962.36 5750.04 4990.25 163.29 0.05 0.04 7924.37 7684.04 6611.20 221.06 0.06 0.06 426.29 426.29 426.29 426.29 426.29 172.54
0.30 0.25 0.28 0.33 1705.54 0.30 0.22 0.30 0.38 106.51 0.26 0.20 0.24 0.30 425.91 0.30 0.25 0.29 0.33 426.46 0.32 0.27 0.30 0.38 426.89 0.30 0.26 0.29 0.33 426.29
Notes: Each year we compute sample average of each variable. In the above table we report the time series mean, quartile1 (Q1), median, quartile3 (Q3), and sample size (N) of each variable average computed over 35 years period. All the dollar amounts are adjusted to 2005 US dollars using Consumer Price Index (CPI; obtained from the Website of Federal Reserve Bank of St. Louis [http://research.stlouisfed.org/]). MVE is market value of equity.
42
TABLE 2 Estimates of Liquidity Risk for Zero and Positive Advertising Portfolios (One-Year Advertising Portfolio - One-Year Estimation Period) Advertising Portfolios Zero Advertising Positive Advertising (Z1) Model 1 Market Risk Liquidity Risk Adj. R2 Model 2 Market Risk Liquidity Risk SMB HML UMD Adj. R2
a
Difference in Risks (Z1) - (Z2) .075 (.65 ) .4 7 9a (2.89)
(Z2) 1 .2 9 6a (18.45) .4 3 4a (4.74) .7 0
1 .3 7 1a (10.06) .9 1 3a (5.47) .4 0
1 .3 2 4a (8.99) .9 6 3a (4.18) .8 0 3a (4.27) -.041 (-.21) -.215 (-1.09) .6 4
1 .1 0 6a (26.25) .3 8 0a (6.18) .9 9 9a (18.70) -.085 (-1.15) -.233a (-4.95) .9 5
.217 (1.57) .5 8 3b (2.46) -.195 (-1.05) .044 (.22 ) .017 (.09 )
denotes p<= .01 denotes p<= .05 denotes p<= .1 (two-sided) t-values are in parentheses, year dummy coefficients not reported Model 1 is liquidity risk augmented CAPM which comprises market factor and Liu's liquidity factor Model 2 is Carhart 4-factor model which comprises market factor (Market), size factor (SMB), book-to-market factor (HML), momentum factor (UMD) and liquidity factor (Liquidity)
(two-sided), b
(two-sided), c
43
TABLE 3 Estimates of Liquidity Risk for Positive Advertising Portfolios (OneYear Advertising Portfolio - One-Year Estimation Period) Advertising Portfolios Lowest Advertising (P1) Model 1 Market Risk Liquidity Risk Adj. R2 Model 2 Market Risk Liquidity Risk SMB HML UMD Adj. R2
a
Highest Advertising (P2) 1 .3 2 4a (14.47) .4 8 9a (4.25) .6 2 (P3) 1 .2 4 5a (17.21) .2 6 6b (2.55) .7 1 (P4) 1 .1 6 1a (26.59) .1 1 2c (1.81) .8 1 Difference in Risks (P1-P2) .1 3 2c (2.02) .3 8 0a (4.81) (P1-P3) .2 1 0a (3.28) .6 0 3a (6.77) (P1-P4) .2 9 5a (3.33) .7 5 7a (6.57)
1 .4 5 6a (14.11) .8 6 9a (6.80) .5 5
1 .1 9 8a (14.85) .9 1 9a (9.59) 1 .1 6 4a (13.90) -.209c (-1.73) -.296a (-3.59) .8 5
1 .0 7 2a (23.29) .4 0 3a (4.62) 1 .1 8 9a (16.39) -.160c (-1.73) -.238a (-3.47) .9 1
1 .0 9 3a (21.14) .2 0 7a (2.73) 1 .0 4 8a (16.27) -.059 (-.79) -.235a (-5.00) .9 4
1 .0 6 2a (32.16) -.010 (-.20) .5 9 3a (11.49) .088 (1.46) -.162a (-3.40) .9 4
.1 2 6c (1.93) .5 1 7a (5.86) -.025 (-.38) -.049 (-.62) -.057 (-.92)
.105 (1.49) .7 1 3a (7.19) .116 (1.27) -.150 (-1.56) -.061 (-.80)
.1 3 6c (1.72) .9 3 0a (10.30) .5 7 1a (7.32) -.297b (-2.57) -.134 (-1.41)
denotes p<= .01 (two-sided), b denotes p<= .05 (two-sided), c denotes p<= .1 (two-sided) t-values are in parentheses, year dummy coefficients not reported Model 1 is liquidity risk augmented CAPM which comprises market factor and Liu's liquidity factor Model 2 is Carhart 4-factor model which comprises market factor, liquidity factor, size factor (SMB), book-to-market factor (HML), and momentum factor (UMD),
44
TABLE 4 Impact of Advertising on Liquidity Risk at the Firm-Level (One-Year Advertising Portfolio - Three-Year Estimation Period) (1971-2005) P1 Intercept Advertising Total Assets Leverage Book to Market Age R2 N Unique Firms
a
P2
a
P3 -.202 (-3.32) (-2.54) -.000 (-.70) .3 3 7a (3.30) .004 (.15 ) -.011 (-1.10) .1 0 8490 1890
a
P4 -.126a (-2.65) .0 0 0bx (2.38) .000 (.45 ) .2 2 4a (2.76) .1 0 7a (5.19) .0 2 9a (4.81) .0 6 10743 1324
.493 (4.68) .134 (.77 ) -.000 (-.28) .219 (1.41)
.026 (.33 ) -.078b (-2.11) -.000c (-1.78) .070 (.53 ) .0 8 9a (2.78) .012 (.79 ) .0 7 7106 2019
-.014b
.0 6 6c (1.73) .009 (.37 ) .0 6 6001 1895
denotes p<= .01 (two-sided), b denotes p<= .05 (two-sided), c denotes p<= .1 (two-sided) xActual coefficient is 0.00047 t-values are in parentheses, year dummy coefficients not reported Advertising and total assets are adjusted to 2005 US dollars using Consumer Price Index
45
Table 5 Impact of Number of Investors on Liquidity Risk at the Firm Level Liquidity Risk .072 (.81 ) Log (No. of Institutional Investors) Log (No. of Individual Investors) Total Assets Leverage Book to Market Age Profit Margin (%) Market Share (%) B 2B B 2C .0 3 9c (1.90) -.021c (-1.80) -.000a (-7.90) .054 (.77 ) .0 3 6b (2.21) -.004 (-.32) -.000a (-2.82) .010(.5 5 )-.045 (-.69) .143 (1.03) .0 3 23013 5212
Intercept
R2 N Unique Firms
a
denotes p <= .01 (two-sided), b denotes p <= .05 (two-sided), c denotes p <= .1 (two-sided) Year dummy coefficients not reported.
46
TABLE 6 Economic Effect of Increase in Advertising on Firm Value via Impact on Cash Flow, Market Risk, Liquidity Level and Liquidity Risk P1 70 .15 61 .70 .2 5 38.10% 35.00% .49% .51% .72% 1 .456 .869 22.25% 15 .61 .0 6 3.09 .000 .869 22.25% .00% .7 2 4.63% -3.187 1 .453 22.23% .09% -.047 -1.18% 4.33% -.05% 22.20% .23% P2 165.32 148.48 1.38 40.50% 35.00% .49% .51% .72% 1 .324 .489 18.15% 30 .00 .3 5 7.42 -.078 .462 17.91% 1.30% 1.73 5.77% -3.187 1 .317 18.10% .25% -.047 -1.18% 4.33% -.05% 18.10% .28% P3 455.91 464.94 6.88 38.84% 35.00% .49% .51% .72% 1 .245 .266 15.73% 71 .72 1.72 23 .25 -.014 .241 15.51% 1.39% 4.75 6.62% -3.187 1 .233 15.66% .46% -.047 -1.18% 4.33% -.05% 15.68% .32% P4 5679.30 5534.03 172.99 38.81% 35.00% .49% .51% .72% 1 .16110 .11211 13.88% 788.50 43 .25 276.70 .00016 .13317 14.06% -1.27% 41 .69 5.29% -3.187 1 .13623 13.73% 1.10% -.047 -1.18% 4.33% -.05% 13.83% .37%
1 2 3 4 5 6 7 8 9
Market Value of Equity (MVE; M$) (Our Table 1) Sales (M$) (Our Table 1) Advertising (M$) (Our Table 1) Gross Margina,b Tax rate Monthly Risk-free Ratea,c Monthly Market Risk Premiuma,ca,d Monthly Liquidity Risk Premium Market Beta (Our Table 3) Liquidity Beta (Our Table 3) Annual Cost of Equity = 12 x {(6) + ((7) x (9)) + ((8) x (10))} Estimated Cash Flows (M$) = (1) x (11) Increase in Advertising (M$) = 25% x (3) Increase in Sales (M$) = .2 x 25% x (2) ? Liquidity Beta (Our Table 4) Liquidity Beta After 25% Incr. in Advt. = (10) + ((13) x (15)) New Cost of Equity = 12 x {(6) + ((7) x (9)) + ((8) x (16))} % ? MVE Due to Liquidity Risk = [(12)/(17) - (1)]/(1) ? Post-Tax Cash flow = {1 - (5)} x {(4) x (14)} % ? MVE Due to Cash Flow = [(19)/(11)] / (1) McAlister et al. Coefficient (Their Table 2 Column 1) New Market Beta = (9) + {(21) x ((13) / (2))} New Cost of Equity = 12 x [(6) + ((7) x (22)) + ((8) x (10))] % ? MVE Due to Mkt. Risk = [(12)/(23) - (1)] / (1) Coefficient of Log Advertising on Log Relative Spread from Grullon et al. (Their Table 5, Column 2) % ? Rel. Spread Due to Increase in Advt. = 25% x (25) Average eRelative Spread for Grullon et. al. (Their Table 4, Panel A) Decrease in Relative Bid-Ask Spread = (26) x (27) New Annual Cost of Equity = (11) + (28) % ? MVE Due to Liquidity Level = [(12) / (29) - (1)] / (1)
12 13 14 15
18 19 20 21 22 24 25 26 27 28 29 30
a b
Average for time period 1971-2005 Computed as (Sales - Cost of Goods Sold - Advertising)/Sales c Monthly values of risk free rate and market risk premium as reported on WRDS d Monthly values of liquidity risk premium from Liu e Average of relatinve bid ask spreads of the 25 portfolios in Panel A
47
Figure 1 - Conceptual model
Product Market Outcomes Market Share and Profit Margin Advertising Expenditure Investor Awareness
Investor Pool No. of Institutional Investors Liquidity Risk No. of Individual Investors
48
FIGURE 2 PORTFOLIO FORMATION AND ESTIMATION
Portfolio formation period = 1 year, Estimation period = 1 year
1972 1973 1974
?? ? ?? ?? ? ?? ? ? ?? ?
34 Estimation Periods
2005
Portfolio formation period = 3 years, Estimation period = 1 year
1974 1975 1976
?? ? ?? ?? ? ?? ? ? ?? ?
32 Estimation Periods
2005
1971
1972 1973 1974 1975 1976 1977
--------------
2001 2002 2003 2004 2005
Portfolio formation period
Estimation period
49
FIGURE 3 DIFFERENCE IN LIQUIDITY RISKS OF ADVERTISING PORTFOLIOS (ONE-YEAR ADVERTISING PORTFOLIO - ONE-YEAR ESTIMATION PERIOD) Liquidity Risk of Zero Advertising Portfolio (Z1) - Liquidity Risk of Positive Advertising Portfolio (Z2)
50
FIGURE 4 DIFFERENCE IN LIQUIDITY RISKS OF ADVERTISING PORTFOLIOS (ONE-YEAR ADVERTISING PORTFOLIO - ONE-YEAR ESTIMATION PERIOD) Liquidity Risk of Portfolio 1 (P1) - Liquidity Risk of Portfolio 2 (P2)
Liquidity Risk of Portfolio 1 (P1) - Liquidity Risk of Portfolio 3 (P3)
Liquidity Risk of Portfolio 1 (P1) - Liquidity Risk of Portfolio 4 (P4)
51
FIGURE 5 DECILE PORTFOLIO LIQUIDITY RISK AND MEDIAN ADVERTISING EXPENDITURES
5
10
15
20
25
30
35
40
383
52
Web Appendix
Liu's Liquidity Risk Factor (LIQ) Liquidity refers to the ease of trading a stock at the market price, at low cost, in sufficient quantities quickly (Liu 2006). The ability to trade a stock at the market price is also called the price impact of a trade. Generally, a trade in a liquid stock does not impact its market price, while an illiquid stock's trade can change the price a lot, making it less likely that an investor can trade his full desired quantity at the market price. In case the price impact is very high, an investor may decide to not trade at all. The bid-ask spread of a stock represents the cost of the trade, with the illiquid stocks having a higher bid-ask spread. Finally, an investor may be unable to find a counter party to trade an illiquid stock. Researchers use various types of data to capture the liquidity of a stock. Most are based on actual trading data - e.g. turnover, bid-ask spread etc. If the measures do not account for the number of zero-trading days, then the measure does not comprehensively capture the notion of ease of trading. Since the liquidity risk measures are derived from the underlying liquidity level measures, the corresponding liquidity risk measure is also likely to be less complete. Liu (2006) first calculates the individual stock liquidity factor LM12, and then uses LM12 to develop the market liquidity factor LIQ.
Liu's liquidity measure (LM12): LM12 is computed for each stock based on the data from the previous 12 months. LM12 Number of zero trading days in prior 12 months 1/(12 month turnover ) * 21*12 Deflator TD
53
The 12-month turnover is the stock's turnover calculated as the sum of daily turnover over the prior 12 months. Daily turnover is the ratio of number of shares traded on a day to the number of shares outstanding at the end of the day. Twenty-one represents the average number of trading days in a month. TD is the total number of the trading days in the market over the prior 12 months. Deflator is a number chosen such that 0
1/(12 month turnover ) Deflator 1, for all sample
stocks. Liu chooses a deflator value of 11,000 for constructing LM12.
Liu's liquidity risk factor (LIQ): The market-wide liquidity factor, LIQ, is constructed using mimicking portfolio approach similar to that of Fama and French (1993). Each year, the stocks are sorted on their LM12 value and grouped into three equal-weighted portfolios. LIQ is computed as the difference between the monthly stock returns of the most illiquid and most liquid portfolios. The values of LIQ are available on a monthly basis.
Comparison with other liquidity factors: Several other liquidity measures have been proposed in the finance literature (Amihud 2002, Pástor and Stambaugh 2003, Sadka 2006). All these measures are based on actual trading data, and hence do not fully account for zero trading days. For example, Amihud (2002) estimates the ratio of daily absolute return to daily dollar trading volume averaged over a 12 month period. If there is no trading, then this ratio cannot be calculated. Pástor and Stambaugh (2003) derive their measure based on daily trading data over a month with a minimum of 16 trading days. If a stock does not trade or trades on less than 16 days in a month, then its liquidity measure cannot be estimated. The LM12 liquidity measure is a
composite measure that captures the inability of the investor to trade when needed. We believe that LM12 is a more appropriate liquidity measure than measures such as the Pastor-Stambaugh or the Sadka
54
measure. The latter two measures primarily capture the price impact of trading, which is only one dimension of liquidity. Thus we feel that Liu's (2006) measure captures more fully the non-trading
risk to an investor.
Accounting for Delisted Firms Many firms delist from stock exchanges each year for various reasons. Shumway(1997) and Shumway and Warther (1999) report that on average 1.2% NYSE and AMEX firms and 5.6% NASDAQ firms delist each year for performance related reasons (see Beaver, McNichols, and Price 2007 for a review). Once a firm delists, it stops trading on the stock exchange and CRSP sets the firm's subsequent stock returns to missing values. If delisting is not accounted for in the portfolio returns calculation, it can introduce a bias in the portfolio returns. For delisted firms, CRSP reports the delisting code and, if available, delisting returns. Delisting code provides the reason for delisting. Delisting return is the return from the firm's delisting date to its first off-exchange trade in the OTC (over the counter) markets. CRSP does its own research to estimate delisting returns. When the delisting return is unavailable, CRSP sets the delisting return to missing. Shumway (1997) shows that ignoring missing delisting returns biases the portfolio returns. Consider a portfolio consisting of two NYSE firms, A and B. Suppose that firm B delists in month t and trades subsequently in OTC markets, and has an actual delisting return of -30%. Example: Actual (equally-weighted) portfolio return
Price Firm A Firm B Portfolio of A and B Month t 100 100 200 Month t+1 Return 110 70 180 10% -30% -10% Month t 100 100 200 Month t+1 110 0 110 Return 10% -100% -45%
55
The actual portfolio return is -10%. But, if the return for firm B is missing on CRSP and is mistakenly coded as -100% by the researcher, this introduces a downward bias in the return to -45%. If the researcher omits this firm from analysis in month t+1, then the portfolio return is biased upwards to 10%, since only stock A is part of the portfolio. Shumway (1997) and Shumway and Warther (1999) obtained the average delisting returns by examining the prices of the firms on OTC (over the counter) markets. We follow Shumway (1997) and Shumway and Warther (1999) to adjust the missing delisting returns. We proceed as follows. If delisting return is not available and reason for delisting is performance related (delisting code between 500 and 599), we follow Shumway (1997) and Shumway and Warther (1999) and set the missing delisting returns to -30% for NYSE and AMEX firms and to -55% for NASDAQ firms in the delisting month. The firm is not considered as part of the portfolio in subsequent months.
56
FOOTNOTES
i
Portfolio theory (e.g., Lintner 1965; Sharpe 1964) posits that investors can diversify away firm-specific risks by investing in a portfolio of stocks. In equilibrium, the risks that are priced in the stock market will be systematic risks. We estimate liquidity risk as the slope coefficient in a regression of returns on a liquidity factor, similar to how the market beta is estimated as the slope coefficient in a regression of returns on the market factor. The exact estimation details are discussed later in the paper.
ii
It is possible that when a buyer or seller has valuable information and wants to trade quickly before other investors (i.e., they desire to trade immediately and demand liquidity), there may be no one to take the opposite side (i.e., supply liquidity). In practice, liquidity can be provided either by investors who want to trade but do not have a need to trade immediately (e.g., because they do not have short lived information) or by professional traders such as market makers on the NYSE and NASDAQ.
iii
Because the number of market makers is not affected in the short term by marketing activities, we do not focus on them. Also, as discussed earlier, rather than providing liquidity, trading by institutional investors tends to use up liquidity. Hence, attracting institutional investors is not likely to reduce liquidity risk.
iv
Several other liquidity risk factors have been proposed in the literature (e.g. Amihud 2002, Pástor and Stambaugh 2003, Sadka 2006). However the liquidity measures used to develop these risk factors use information from actual trades and do not include any data which measures how often there is no trading at all. Thus their liquidity risk measures do not fully capture an investors' difficulty of trading.
v
We also estimate the relationship between advertising and the number of individual and institutional investors for the period 1997-2005, using fixed effect panel model with AR(1) serial correlation in errors. We control for product market impacts (market share, profitability, BtoB and BtoC) as well as other variables suggested in Grullon, Kanatas and Weston (2004). We find that advertising positively impacts only the number of individual investors. Results are not separately tabulated, but are available upon request.
vi
Whenever available, we use the historical SIC code from CRSP to classify firms into the Fama French industries. If missing, we use the current SIC code as reported on COMPUSTAT (Hanka 1998).
vii
We have derived our estimates assuming cash flow neutrality, i.e., the incremental advertising cost is offset by increases in sales. We caution that if a firm desires to increase advertising purely to decrease cost of capital without any change in sales, this may not be an economically viable decision. Individual firm impacts will depend on their advertising sales elasticity.
57
doc_724633196.docx
In finance, liquidity risk is the risk that a given security or asset cannot be traded quickly enough in the market to prevent a loss (or make the required profit).
Study of Advertising on Liquidity Risk
Abstract Marketing actions can impact firm value through both product and capital markets. Recent literature in finance suggests that in addition to systematic market risk, liquidity risk the non-diversifiable systematic risk that a firm's stock may become illiquid in times of market stress - is also priced. The authors examine a large panel of more than 1,800 firms from 1971 to 2005 and show that advertising lowers liquidity risk by increasing the number of individual investors in a firm. The impact of higher advertising on liquidity risk is more pronounced for firms that are younger, operate in BtoC markets, and have advertising expenditures between about $ .5 million and $ 20 million. Simulations show that increasing advertising expenditures by 25% increases firm value by up to 1.4% just due to a reduction in liquidity risk. Overall, the results suggest that, in addition to the product market impacts, researchers and managers should consider the valuation impact of marketing activities via their effect on investors in capital markets. Key words: Advertising, Shareholder value, Liquidity risk
1
The marketing profession has demonstrated the positive effect of marketing actions on shareholder value, either by improving future cash flows or by reducing their riskiness (Srinivasan and Hanssens 2009). Several earlier studies have examined the influence of marketing on a firm's cash flow (e.g., Mizik and Jacobson 2007, Fornell et. al 2006). More recently researchers have studied the impact of marketing activities on firm risk, using various risk metrics. A partial list includes the effect of advertising (McAlister, Srinivasan and Kim 2007) and customer satisfaction (Tuli and Bharadwaj 2009) on systematic market risk; consumer negative voice (Luo 2007), CSR actions (Luo and Bhattacharya 2009), and innovation (Sorescu and Spanjol 2008) on idiosyncratic firm risk; customer satisfaction (Gruca and Rego 2005) on cash flow variability; and customer relationships on sales volatility (Tuli, Bharadwaj, Kohli 2010). Investigating the impact of marketing activities on risk is clearly an important topic for marketers to study. In this paper, we examine whether advertising by a firm impacts liquidity risk, a systematic risk that has not been explored in the extant marketing literature. Recent literature in finance suggests that in addition to the commonly understood systematic market risk, investors are also concerned about systematic liquidity risk. Liquidity represents the ease with which an investor is able to trade a stock at the market price, at low cost, in sufficient quantities quickly. Liquidity varies over time both for individual stocks and for the market as a whole. Liquidity risk represents the non-diversifiable systematic risk that a firm's stock may become illiquid in times of market stress. For example, during market downturns when the market declines in value or overall liquidity dries up, investors may be unable to sell some of their assets quickly to meet consumption or other needs. Hence, investors would be willing to pay a premium and accept lower returns on a stock that is expected to be liquid in down or illiquid markets due to its low covariance with the market. If a stock has low liquidity
2
risk, investors require a smaller risk premium to compensate for this systematic risk, thus decreasing the firm's cost of capital. This decrease in cost of capital will increase firm value, all else equal.i Acharya and Pedersen (2005) find that the cost of equity is lower by about 1.1 % per year for stocks with low liquidity risk. Liquidity risk is shown to be priced not only in the US but also across a large number of global markets, thereby suggesting that it is a pervasive risk which is important in determining firm value (Lee 2010). In this paper, we explore the role of marketing, and more specifically advertising, in reducing liquidity risk through its impact on investor behavior in capital markets. We draw upon the prior literature in finance and hypothesize that a firm may be able to reduce its liquidity risk by increasing the number of individual investors. Both theoretical and empirical results in the finance literature suggest that improved awareness can increase the number of investors. Merton (1987) asserts that investors are likely to invest only in firms they know about. Consistent with this view, Huberman (2001) finds that investors are more likely to hold shares of local firms that they are aware of, and concludes that "people invest in the familiar while often ignoring the principles of portfolio theory.? Since advertising is one of the main instruments available to marketing managers to increase awareness, we examine its role in reducing liquidity risk. Anecdotal evidence suggests that managers realize the importance of improving individual investor awareness. For example, Federated Department Stores changed its corporate name to Macy's, and stated that ?by aligning our corporate name with our largest brand, we will increase the visibility of the company with customers, leverage the world-famous Macy's brand name, and get more credit for our accomplishments in the marketplace? (Lundgren 2007; italics added). Similarly, Sun Microsystems changed its trading symbol from SUNW to
3
JAVA to make the firm more visible to new investors (Jonathan Schwartz, President and CEO, 2007). Our analysis proceeds along four related lines of inquiry. First, we compare the liquidity risk of firms that do not advertise with firms that report advertising expenses. Using data on more than 1,800 publicly traded US firms over a thirty-five year period (1971 to 2005); we find that firms with no advertising are subject to significantly higher liquidity risk than firms that advertise. We obtain qualitatively similar results when we use size-matched firms, suggesting that our results are not an artifact of firm size. Second, we investigate whether the level of advertising affects the level of liquidity risk. We document that liquidity risk declines monotonically across quartiles when firms are grouped annually based on the reported advertising expense. The results are qualitatively similar when we use alternative model specifications; therefore we conclude that the level of advertising is highly correlated with the level of liquidity risk. Third, we show that a firm's liquidity risk is negatively related to the number of individual investors. We also show that advertising is more effective in reducing liquidity risk in BtoC (compared to BtoB) firms as well as in younger firms. Taken together, these results support the notion that product advertising improves a firm's awareness among individual investors in capital markets and lowers liquidity risk. Finally, we assess the economic significance of this result by simulating the valuation impact of potentially increasing advertising expenditures by 25%. We find that firms in the highest advertising quartile (mean annual advertising $173 million) do not see any further decreases in their liquidity risks, possibly because these firms are already well known to individual investors and have low levels of liquidity risk. Firms in the lowest quartile of advertising expenditures also do not see any benefits, because the absolute amounts of
4
advertising would still be extremely low to affect investor recognition appreciably, even after a 25% increase in the advertising budget. The main impact is felt by firms in the middle two quartiles with moderate levels of advertising expenditures, whose market value can increase by up to 1.4 % solely due to this liquidity risk effect. As expected, the cash flow impact of advertising through increased sales results in the largest change in value. However, the change in value via lower liquidity risk for firms in the middle two quartiles is about 20% of the cash flow impact on value, indicating that understanding the effect of advertising on liquidity risk and firm value is important. The results in this paper make significant contributions to the increasingly important literature studying the impact of marketing actions on firm value. First, we introduce liquidity risk as a hitherto unexplored (in the marketing literature) but important driver of firm value and suggest that marketing efforts such as product advertising can lower liquidity risk by affecting investor perception and behavior. Second, we also show that liquidity risk is negatively related to the number of individual investors. Joshi and Hanssens (2010) find that advertising has a direct effect on firm value through the spillover and signaling effects on investors. Their discussion implies that the driving force for these effects is the product market effect of advertising in making the stock more attractive to investors. Our research complements and extends the extant literature by identifying a capital market effect of advertising. The impact of marketing activities on liquidity risk has, to the best of our knowledge, not been examined in the marketing or the finance literature. Third, our results also provide greater insight into the economic significance of liquidity risk by simulating the impact of increased advertising on firm value, both through increased cash flow and reduced liquidity risk (which reduces the cost of equity capital). Finally, at a broader level, our finding that advertising lowers liquidity risk and leads to a lower cost of
5
capital may enable a firm to convert hitherto unprofitable projects (e.g., a brand extension decision, new product introduction) into viable ones. Consequently, our results indicate that the returns to advertising have likely been underestimated in the literature. The rest of the paper is organized as follows. We first provide the conceptual background and discuss how marketing actions such as advertising can lower liquidity risk. We then describe the data and the methodology used to estimate liquidity risk and discuss the main results. Next, we estimate the incremental value created by advertising via its impact on liquidity risk and conclude by discussing the managerial implications of our findings.
Advertising and Liquidity Risk
The most widely accepted theoretical model for asset pricing in the finance literature is the Capital Asset Pricing Model (CAPM), where all investors are assumed to hold a combination of a riskless asset (such as a treasury bill) and a diversified market portfolio. The allocation between these two assets reflects the investor's choice of how much risk to take. In the CAPM, the risk of a stock, popularly known as the ?market beta?, is the standardized covariance between the stock return and the market risk factor. Firm-specific risks are diversified away by the investor. However, the CAPM assumes a perfect world in which there are no transactions costs; thus, investors can diversify costlessly. In such a perfect world, the only risk that explains crosssectional stock returns is the market beta. Recent research in finance argues that in practice, the costless trading assumption of CAPM does not hold thereby suggesting that investors are also concerned about the liquidity of their portfolio.
6
Not all securities in the US stock markets trade every day. For example, of the about 5,000 stocks available in CRSP, only 80% traded on each of the approximately 250 trading days in 2007. About 5% of the stocks did not trade on 40 days or more. The lack of trading arises due to several transactions costs that reduce liquidity. These costs include (i) out-of-pocket costs such as brokerage fees, order-processing costs, etc., and (ii) opportunity costs incurred due to investors' inability to find suitable counterparties to trade with immediately. A firm with lower transaction costs is characterized as having a higher liquidity level, which is generally described as the ease of trading a stock at the market price, at low cost, in sufficient quantities quickly (Lin, Singh, and Yu 2009, Liu 2006). Investors will be concerned about liquidity and will take the entire future stream of these transaction costs into account when valuing a stock. Studies in finance have shown that investors are compensated with higher returns for investing in lessliquid assets (e.g., Amihud and Mendelson 1986). Liquidity also varies over time, both for individual stocks as well as in the overall market. This suggests that investors holding a portfolio of assets at any point of time would be concerned both with its current and future liquidity when making their investment decisions. For example, as investment opportunities change in the marketplace, investors may need to trade and readjust their portfolios. In addition, higher future consumption needs (e.g., to buy a home or to finance a child's education) may require them to liquidate some of the assets in their portfolio. Thus, in addition to the market risk, investors are also concerned with the liquidity risk of their portfolio. Pastor and Stambaugh (2003), Acharya and Pedersen (2005), Sadka (2006), and Lee (2010), among others, document that liquidity risk is an important risk factor that affects stock returns. The preceding discussion suggests that when evaluating individual stocks in their portfolio, investors would be concerned about the systematic component of a stock's liquidity
7
risk, analogous to the notion of systematic market risk. The firm-specific component would be diversified away and would not be relevant for the investor. Liquidity risk represents the nondiversifiable systematic risk that a firm's stock may become illiquid in times of market stress. Investors are compensated with a risk premium for this liquidity risk, arising due to commonality or covariation in liquidity over time (e.g., Chordia, Roll, and Subrahmanyam 2000, especially the discussion in their Section 1.2). Hence, an investor would require a smaller risk premium for a stock with low liquidity risk that contributes to reducing the overall liquidity risk of her portfolio. This lower risk premium will decrease a firm's cost of capital and increase firm value. We emphasize that liquidity level and liquidity risk represent two distinct, albeit related, aspects of liquidity that are priced by investors. Advertising can potentially impact both the liquidity level (a positive link is documented in Grullon, Kanatas and Weston (2004)) and liquidity risk of a firm. We focus on the hitherto unexplored impact of advertising on liquidity risk and present our conceptual model in Figure 1. Insert Figure 1 about here Advertising can attract investors to a firm in two ways. First, advertising can impact customer behavior and increase sales, market share, or profits of firms (Vakratsas and Ambler 1999). This makes the firm more attractive as an investment. We label this as the product market outcome of advertising. Second, media advertising intended to create awareness among consumers in general can also have a spillover impact on the investor community, resulting in higher overall awareness about the firm and its products (Lovett and MacDonald 2005, Joshi and Hanssens 2010). Traditionally, investors have been classified into two broad groups - individual investors and institutional investors such as mutual funds, hedge funds, and pension funds. Institutional investors are typically already aware of all investment opportunities, and thus they
8
may be attracted to higher advertising firms mainly due to the implied positive product market outcomes, and not due to the spillover effect of advertising. On the other hand, individual investors face time and capital constraints, and are much more likely to be affected by both increased awareness and product market signals of advertising. For example, Frieder and Subrahmanyam (2005) conclude that individual investors prefer to invest in stocks with easily recognized brands and Fehle, Tsyplakov and Zdorovtsov (2005) document a significant increase in buying activity by small investors immediately after Super Bowl advertisements. Thus, advertising likely creates awareness of and preference for the firm and helps increase the consideration set (Mitra and Lynch 1995) in which individual investors evaluate investment opportunities. Research in finance suggests that trading by individual and institutional investors affect liquidity risk differently. Trades occur when buyers and sellers exchange assets at the ongoing market price, which makes the stock liquid.ii While active institutional investors exhibit herd behavior and tend to use up liquidity by typically trading in the same direction (Dennis and Strickland 2002), individual investors step in to provide liquidity (Kaniel, Saar and Titman 2008). Thus we conjecture that a firm that has a larger pool of individual investors whose trading provides liquidity on a consistent basis will have lower liquidity risk. Investors in such stocks can be more confident that the presence of a large pool of liquidity providers will enable them to complete their trades when needed. Simply put, investors will require a smaller risk premium as compensation for having such stocks in their portfolio. If advertising is to reduce a stock's liquidity risk, the effect has to be to attract liquidity providers to the stock and ensure that the stock's liquidity does not dry up, especially in times of market stress. This can be achieved if advertising is able to attract a large number of individual
9
investors whose ability to participate in the stock market does not fluctuate greatly in good times and bad times (as discussed earlier), thus providing an ex-ante expectation of a steady source of liquidity, and lowering liquidity risk.iii This suggests that firms that advertise will have a lower liquidity risk than firms that do not. Furthermore, the level of advertising will be related to the level of liquidity risk. All else equal, firms that advertise more heavily will be able to create awareness among a larger pool of individual investors and will reduce liquidity risk to a greater extent than firms that have limited advertising budgets. In other words, as the magnitude of advertising expenditures increases, the individual investor pool will be larger, leading to lower liquidity risk. We test the following hypotheses in this paper: H1a: Firms with zero advertising expenditure will have higher liquidity risk than firms that advertise. H1b: Firms that spend more (less) in total advertising dollars will have lower (higher) liquidity risk. H2: Liquidity risk will be lower for firms that have a larger number of individual investors.
Data and Methodology
Sample Selection We use data for US based firms available in the merged CRSP COMPUSTAT file from Wharton Research Data Services (WRDS) during the 35 years from 1971 through 2005. We limit our analysis to this period as only a small number of firms reported advertising expenditure
10
before 1971 and the liquidity risk factor LIQ (Liu 2006), obtained from the author, is only available to us till 2005. We exclude foreign firms that are traded as American Depository Receipts (ADRs), closed end funds and other securities, retaining only common stocks of domestic US firms with a share code of 10 or 11. On average, about 62% of the firms have missing advertising data per year. Because we cannot reliably determine whether these firms actually had significant advertising expenditure but failed to report it separately or they did not spend any material amount on advertising, we exclude these firms from the sample. Finally, we retain only those firms in our sample that have monthly stock returns data available on CRSP. This results in an average sample size of 1,812 firms per year (63,429 firm years) representing about 30% of the COMPUSTAT firms per year. Our model estimation requires monthly data on stock returns, returns on U.S Treasury bills (the risk free rate), data on the three factors proposed by Fama and French (1993) and the momentum factor proposed by Carhart (1997); these data are all obtained from WRDS. All financial statement data are from COMPUSTAT, stock returns data are from CRSP, and ownership data are from the Thomson Reuters database. In comparison to other studies in marketing using financial data, our sample is larger in size, spans a longer time period of thirty-five years, and is broader in scope including both large and small firms. For example, McAlister, Srinivasan, and Kim (2007) analyze 644 firms over a period of 22 years (1979-2001). They limit their sample to the firms listed on New York Stock Exchange (NYSE). In contrast, our sample includes firms listed on any of the three stock exchanges - NYSE, AMEX, and NASDAQ.
11
Methodology to Estimate Liquidity Risk We use a procedure similar to the estimation of systematic risks in the Fama French model (Fama and French 1993). Typically, a risk factor is created as the difference in returns between stocks with high and low values of the variable of interest (e.g., firm size, book to market ratio, etc.). The risk level is then estimated as the covariance of a single stock's returns with the risk factor. We follow this approach and use the liquidity risk factor LIQ (Liu 2006). Liu (2006) uses his LM12 liquidity level measure to sort all firms in the market into three liquidity groups. LIQ is then estimated as the difference in returns between portfolios of stocks with low and high liquidity. Details of the measures are provided in the Web Appendix. The liquidity risk is estimated as the slope coefficient of the regression of a stock's monthly excess returns on the liquidity risk factor LIQ (and other risk factors as discussed later). Others (e.g., Eckbo and Norli 2005) use a similar approach to measure liquidity risk, except that they use turnover (number of shares traded to number of shares outstanding) as a measure of stock's liquidity level. We prefer Liu's (2006) liquidity level measure LM12 because it additionally captures the difficulty of trading (LM12 uses both the number of zero trading days and turnover); thus the LIQ measure is more comprehensive.iv In order to estimate the link between advertising and liquidity risk, we use the calendartime portfolio method (Srinivasan and Hanssens 2009). This method is commonly utilized in the finance literature and has also been used in prior studies in marketing (e.g., Sorescu, Shankar and Kushwaha 2007). Using a portfolio of firms rather than individual firms increases the precision of the estimated coefficients (Beaver, Kettler and Scholes 1970). We first rank all the stocks based on their advertising expenditures during the portfolio formation period of one year. We then either divide all the stocks into two portfolios (zero-advertising and positive-advertising, to
12
test H1a), or divide all stocks with positive advertising expenses into four equal sized portfolios (to test H1b). We reassign firms to portfolios at the end of every portfolio formation period and repeat this process over 1971-2005. We invest equal amounts in all stocks in a given portfolio during the model estimation period of one year and calculate the monthly return on the portfolio. We thus generate a time-series of monthly portfolio returns for every portfolio. In calculating the portfolio return, we also account for firms that delist from the stock exchange over time. A delisted firm could have gone bankrupt, or moved to a different exchange. If a delisted firm's return is not properly accounted for in estimating the portfolio returns, it introduces a bias in the returns. The details of this adjustment are provided in the Web Appendix. We illustrate this calendar-time portfolio procedure for firms with positive advertising expenses. We start by sorting the firms into quartiles, P1 to P4, based on the advertising expenditures in 1971. P1 (P4) consists of all the firms in the lowest (highest) advertising quartile. The corresponding model estimation period is 1972. The equal-weighted portfolio returns are calculated for every month in 1972 using the monthly returns for individual stocks in the portfolio. This procedure yields 12 monthly returns for each of the P1 to P4 portfolios, which are used to estimate the regression models (described below) for each portfolio. We repeat this procedure for the next portfolio formation period, 1972 andreassign firms to the four P1 to P4 portfolios, based on advertising expenses in 1972. Next, we calculate the monthly portfolio returns and estimate the regression model using the 12 monthly portfolio returns in the model estimation period 1973. We continue this procedure until 2005 and end up with 34 nonoverlapping model estimation periods, i.e., 1972 to 2005. As shown in Table 2, each portfolio, on average, consists of 426 firms per year. Figure 2 shows the portfolio formation and model
13
estimation periods discussed above as well as the ones used in the robustness checks discussed later. Insert Figure 2 about here In the first model (liquidity risk-augmented CAPM), we include the market risk factor and the liquidity risk factor, LIQ, as factors explaining expected return (Equation 1). We include these two risk factors because they have strong theoretical underpinnings (e.g., Sharpe 1964, Lintner 1965 for the market risk factor; Pastor and Stambaugh 2003, Acharya and Pedersen 2005, Korajczyk and Sadka 2008 for liquidity risk factor). However, research in finance and marketing (e.g. Srinivasan and Hanssens 2009) also uses four factors - market, size, book-tomarket, and momentum risk factors (Fama and French 1993; Carhart 1997) to capture expected return, even though Fama and French (2004) argue that the last three risk factors are ?brute force? constructs, with little theoretical justification. Hence, we estimate a second model (2) that includes these additional risk factors, as a robustness check. (1) Rpt - Rft = ?p + ?mpMKTRFt + ?lpLIQt + ?pt (2) Rpt - Rft = ?p + ?mpMKTRFt + ?lpLIQt + sp SMBt + hpHMLt + upUMDt + ?pt where Rpt is the return on portfolio p during month t and Rft is the return on a U.S. Treasury bill during the same period; MKTRFt is the difference in return between the risky market portfolio Rmt and the risk free return Rmt in month t, LIQt is return difference between the stocks in the low and high liquidity portfolios in month t, SMBt is the size risk factor in month t and is calculated as the difference between the returns of a portfolio of small vs. large firms, HMLt is the difference in returns of a portfolio of high and low book-to-market stocks, UMDt is the difference in returns between a portfolio of winner stocks with high prior returns and loser stocks
14
with low prior returns, , and ?pt is an error term. The main variable of interest is the liquidity risk, the estimated slope coefficient ?lp. We annually estimate both models for each estimation period from 1972 to 2005 and report the mean of these parameter estimates. Our regressions follow a procedure similar to that suggested by Fama and MacBeth (1973). The model estimation periods are always nonoverlapping periods, to minimize correlation in the estimated betas. Thus, we expect that the parameters estimated in one time period are independent of the parameters estimated in other time periods. In addition, if markets efficiently incorporate the available information into stock prices, monthly stock returns would be serially uncorrelated. Insert Table 1 about here Summary Statistics The first two panels in Table 1 report the summary statistics for firms that advertise (Panel A) and firms that do not (Panel B). We first compute the averages of all the financial variables across all the firms for each year and each portfolio, resulting in 35 mean values for each variable, and report the descriptive statistics of these 35 values. All dollar values are adjusted for inflation and reported in 2005 dollars. On average 1,706 firms reported positive advertising expenses each year. The mean (median) market value of equity is $1,592 ($1,071) million. The mean (median) advertising expenditure is substantial at $45 ($42) million. These firms, on average, spent 7% of their sales on advertising every year and have a mean age of 11.84 years. The sample in Panel B, with an average of 107 firms per year that reported zero advertising, is considerably smaller than that in Panel A. We find that these firms are smaller with a mean (median) market value of equity of $295 ($139) million. Although the mean total assets are $1,120 million, the median is just $211 million and the mean age is 6.75 years.
15
The next four panels in Table 1 report similar statistics for four quartile portfolios formed by sorting firms on their annual advertising expenditures. Each portfolio consists of an average of 426 firms every year. Portfolio P1 includes firms with the lowest amount of advertising expenses. The mean (median) annual advertising expenditure is only $.25 ($.21) million. The mean (median) annual advertising expenditure is $1.38 ($1.17) million for Portfolio P2, the second quartile, and is $6.88 ($6.47) million for portfolio P3. Finally, for the highest advertising quartile, portfolio P4, the mean (median) annual advertising expenditure is $172.99 ($163.29) million. For brevity, in the rest of this section we discuss summary statistics only for portfolio P1 (lowest advertising) and portfolio P4 (highest advertising). The firms in P1 are significantly smaller than the firms in P4. The average market value of equity of P1 is $70 million and is about 1.2% of the mean market value of equity of the firms in portfolio P4. We find a similar pattern in total assets, sales, and firm age, indicating that the firms in P1 are much smaller and younger than the firms in P4. In spite of the size differences, the average advertising-to-sales ratio for both the portfolios is about 7%, suggesting that firms in both portfolios have equal advertising intensity. However, given the small absolute level of annual advertising expenditures in P1 ($.25 million), any visibility generated due to advertising will likely be very small. On the other hand, with the same advertising intensity, the firms in P4 are likely to have a very high visibility due to their much higher level of average advertising expenditures ($172.99 million).
Results
In this section, we begin the empirical analysis by presenting the results of regressions
16
using the liquidity risk-augmented CAPM model (Model 1) and the liquidity risk-augmented Carhart 4-factor model (Model 2). We mainly discuss the results relating to liquidity risk (coefficient of the liquidity risk factor) because the other risk-factors are used as controls for other systematic risks. We also report the results of several robustness checks that we conducted to assess the reliability of our results. Insert Table 2 and Figure 3 about here
Comparison of Zero and Positive-Advertising Portfolios (Hypothesis H1a) In Table 2, we report the estimates of the two models for zero- and positive-advertising portfolios using a one-year portfolio formation and one-year estimation period. The stocks are sorted into two portfolios every year based on the advertising expense (zero and positive advertising) and the regression models are estimated using returns over the subsequent 12 months. There are 34 such model-estimation periods and Table 2 shows the mean and t-values of the betas across these 34 regressions. Assuming that the betas are stationary and serially uncorrelated, we can use a standard t-test to check the significance of the betas, and a paired ttest to test for differences of the betas between portfolios (Litzenberger and Ramaswamy 1979). In both models, liquidity risk is statistically significant both for firms with zero advertising and for firms that report positive advertising expenses. However, consistent with hypothesis H1a, we find that firms that do not advertise are subject to significantly higher liquidity risk than firms that use advertising. Specifically, using Model 1, liquidity risk for firms in the positiveadvertising portfolio averages .43, which is significantly lower than the mean of .91 for firms in the zero-advertising portfolio (p< .01). The results using Model 2 are similar. The corresponding estimates of liquidity risk are .38 and .96, and the mean difference is statistically significant (p<
17
.05). Figure 3 shows the difference in liquidity risk (beta for the zero-advertising portfolio - beta for the positive-advertising portfolio) for Model 1 for each year. The difference is positive in most of the years, as expected. These results indicate that the firms that do not advertise are subject to higher liquidity risk, and support hypothesis H1a. We conduct three additional tests to assess the robustness of our results. First, we extend the portfolio formation period from one to three years. We assign firms to portfolios based on the average advertising expense over the prior three years rather than just in the prior year. This is more likely to reflect the firm's longer term advertising strategy. Further, when we use a oneyear portfolio formation period, we are assuming that most of the effect of advertising on investors occurs during the next year. However if advertising has a longer term effect, then the impact on investors in period t+1 could be due to advertising expenditures in periods t, t-1, t-2 etc. The three-year portfolio formation period accounts for this longer term effect of advertising. We estimate the models in the 32 one-year estimation periods. The results are similar to those reported earlier (difference in liquidity risk; Model 1, .472, p< .05; Model 2, .649, p=.01). Second, we control for firm size because large firms tend to spend more on advertising as compared to small firms. One possible objection to forming portfolios based on advertising expenditure is that it could be equivalent to sorting them on firm size. In addition, larger firms may be more visible to investors as they are more likely to be covered by financial analysts and media. Although we control for size effects in model 2 by including the size factor SMB, we conduct an additional test to ensure that the differences in liquidity risks are primarily due to differences in advertising expenditure. For each firm in the zero-advertising sample, we select a matching firm of similar size from the positive-advertising sample that has total assets within 70130% of the total assets of zero-advertising firm (similar cutoffs are used in Barber and Lyon
18
(1997)). Using a one-year portfolio formation period and one-year estimation period, we continue to find that the liquidity risk for the zero-advertising portfolio is higher than that for the positive-advertising portfolio (difference in liquidity risk; Model 1, .276, p= .077; Model 2, .449, p<.05). We obtain qualitatively similar results where we use a three-year-portfolio formation one-year estimation window (difference in liquidity risk; Model 1, .265, p=.246; Model 2, .924, p= 0.05). This is a stronger test of hypothesis H1a, suggesting that even after explicitly controlling for firm size, advertising reduces liquidity risk. Finally, we pool all the monthly observations to obtain more precise estimates of liquidity risk, and test for differences between the liquidity risk of the two portfolios. For the one-year portfolio formation period, we pool 408 monthly returns and find that the differences are statistically significant (Model 1, .395, p<.01; Model 2, .309, p<.01). For the three-year portfolio formation method we pool 384 monthly returns and again find statistically significant differences (Model 1, .465, p< .01; Model 2, .362, p= .01). Using size matched firms, the pooled sample of both one year portfolio formation (Model 1, .282, p<.01; Model 2, .337, p<.01) and three-year formation period (Model 1, .239, p< .1; Model 2, .294, p< .1) shows significant differences. Overall, the results so far strongly support H1a that firms with positive advertising have lower liquidity risks. In the next section we test whether the amount of advertising is linked monotonically to liquidity risk (H1b). We expect that as firms advertise more, they will have a larger individual investor pool and thus lower liquidity risk.
Comparison of the Four Positive-Advertising Quartile Portfolios (Hypothesis H1b) In Table 3 we report the estimates of the models for the four positive-advertising portfolios (P1 to P4). The first four columns show the values of the risk coefficients for the four
19
portfolios. Two important observations about the estimates of liquidity risk emerge. First, using both estimation models, the liquidity risk of the low advertising portfolio P1 is always the highest and it decreases monotonically as we move from the lowest advertising to the highest advertising portfolio. This observation strongly supports hypothesis H1b, that higher levels of advertising are associated with lower liquidity risk. Second, the liquidity risk of P4 is marginally significant in model 1 (p< .1) and is not significantly different from zero in model 2. Firms spending large amounts on advertising are very visible to investors and will have large individual investor pools, and thus face very low or no liquidity risk. The last three columns in the table show that the liquidity risk of P1 firms is significantly higher than the liquidity risk of firms in portfolios P2, P3, and P4. For example, for model 1 the mean differences in the liquidity risk for P1-P2, P1-P3, and P1-P4 are respectively .380 (p< .01), .603 (p< .01) and .757 (p< .01), and are highly significant. In Figure 4 we plot the differences in liquidity risk (P1 - P2), (P1 - P3), and (P1 - P4) for Model 1 and confirm that the differences are positive in the overwhelming majority of the years. The consistently positive differences between liquidity risk of the lowest (P1) and higher advertising portfolios (P2, P3, and P4) indicate that our results are robust and not restricted to a few years. Insert Table 3 and Figure 4 about here We conduct additional robustness checks similar to those discussed earlier for zero and positive advertising portfolio comparisons. We use a three-year portfolio formation period, sizematched portfolios, and a pooled regression. We find that our results generally hold. The differences in liquidity risk are insignificant only in some of the size matched portfolios using one or three year portfolio formation periods. However, the majority of the differences in
20
liquidity risk (76%) are both in the correct direction and statistically significant at the 10% level or better, providing strong evidence in support of hypothesis H1b. Decile Portfolio Analysis As documented in Table 3, the liquidity risk of portfolio P4 stocks (with an average advertising expenditure of $173 million) is close to zero. In order to more closely examine the level of advertising at which liquidity risk becomes small, we repeat the analysis using one-year portfolio formation and one-year estimation windows by creating deciles rather than quartiles based on advertising expenditures. The results confirm that liquidity risk decreases monotonically with increasing advertising expenditures. We plot liquidity risk against median advertising expenditure for each of the ten portfolios (Figure 5), and find that liquidity risk drops sharply as advertising expenditures increases to about $7 million. Subsequent increases in advertising (until about $40 million) only gradually reduce liquidity risk. Increasing advertising beyond that does not lower liquidity risk by much. However, firms may advertise beyond this level with a view to increasing consumer awareness in product markets. Insert Figure 5 about here Overall, the results in this section strongly support hypotheses H1a and H1b, indicating that as the level of advertising increases, the liquidity risk decreases.
Firm Level Analysis The analysis thus far has used portfolios of stocks to document that advertising is related to liquidity risk. However, as the stocks are grouped into portfolios, we cannot use crosssectional firm level regression models to explore the relationship between advertising, number of investors and liquidity risk. In addition, firm level regressions will allow us to control for a
21
number of firm characteristics that may be driving cross-sectional differences in firm liquidity risk. We also use these regressions in performing economic impact analysis later in the paper. We adopt a two-step methodology used in earlier research (e.g., McAlister, Srinivasan and Kim 2007) to link advertising expenditure with liquidity risk for each firm from 1971 to 2005. To illustrate, for a given year say 1973, we regress the monthly excess return for each firm during the previous 36 months (1971-73) on the market and liquidity risk factors (liquidity risk augmented CAPM, eq. (1)). This generates a slope coefficient on the liquidity risk factor LIQ as an estimate of liquidity risk for the firm. In the second step, we regress the firm-specific liquidity risk on the firm's advertising; and include total assets, leverage, book-to-market, age, and year dummies, as control variables. Every year, we reassign firms to four advertising portfolios based on the advertising in the prior year (e.g., 1972) and re-estimate the regression models. All independent variables are lagged by one year. Preliminary tests on the residuals from our model indicate significant autocorrelation in the errors. Therefore, we use a fixed-effect panel model with AR(1) serial error correlation. The results of the second stage regression linking liquidity risk to advertising are shown in Table 4. Advertising is significantly related to liquidity risk for firms in three of the four portfolios P2, P3 and P4. Advertising has no impact on liquidity risk for firms in P1, the lowest advertising portfolio, where the coefficient on advertising is not statistically significant. The highest impact of advertising is in P2 (?=-.078, p <.05). The impact reduces in P3 (?=-.014, p<.05). We expected the coefficient in P4 to be zero as the investor awareness about these firms should be high and advertising should not impact liquidity risk. We find that the coefficient on advertising is positive and significant (?=.00047, p<.05), even though the economic magnitude of the coefficient is small.
22
Insert Table 4 about here Overall, the results in this section indicate that if the advertising levels are either too low or too high, advertising does not reduce liquidity risk. However in P2 and P3, as the level of advertising increases, liquidity risk decreases. This is likely to occur if higher advertising results in a larger numbers of individual investors. In the next section, we present evidence to test hypothesis H2, and examine whether the number of individual investors in a firm is negatively linked to liquidity risk.
Advertising, Ownership Structure and Liquidity Risk (Hypothesis H2) We examine whether an increased presence of individual investors in a firm is associated with lower liquidity risk as a direct test of hypothesis H2. We have discussed prior research (e.g., Kaniel, Saar, and Titman 2008) which documents that while institutional investors demand immediacy of trade and use up liquidity, individual investors step in and provide liquidity. Thus firms that have a larger number of individual investors would have lower liquidity risk (H2). As before, we follow a two step procedure to test this hypothesis. In the first step, we estimate the annual liquidity risk for each firm using the liquidity risk augmented CAPM over the previous 36 months. We then use these firm level betas (the liquidity risk estimates) as dependent variables in second stage regressions where the main explanatory variable of interest is the number of individual investors. The data on number of institutional investors are only available to us from 1997 and the tests in this section use this subset of data. The number of institutional investors in a given year is obtained as an average of the quarterly reported numbers. Thomson Reuters does not report the number of individual investors. We use the difference between the number of total shareholders
23
(from COMPUSTAT) and the number of institutional investors as an estimate of the number of individual investors for a firm-year. We regress our liquidity risk estimate on the log of the individual investors and include other control variables, viz., log of number of institutional investors, total assets, book-to- market, leverage, firm age, profit margin, market share, dummy variables for BtoB or BtoC firms, and year dummies. We have a total of 23,013 firm-year observations from 5,212 unique firms with non-missing observations on all the model variables from 1997 to 2005. All independent variables are lagged by one year. Preliminary tests on the residuals from our model indicate significant autocorrelation in the errors. Therefore, we use a fixed-effect panel model with AR(1) serial error correlation. The results are shown in Table 5. We find that the liquidity risk of firms is significantly negatively related to the number of individual investors, lending support to H2. Interestingly, we find that liquidity risk is significantly positively related to the number of institutional investors. The opposite signs on individual and institutional ownership suggest that only individual ownership is associated with lower liquidity risk, consistent with the assertion in the finance literature.v Insert Table 5 about here We conducted three further tests to examine whether the effectiveness of advertising in lowering liquidity risk occurs via creating awareness among individual investors (these results are not tabulated separately, but are available from the authors upon request). First, if advertising is effective in reducing liquidity risk for a firm due to its ability to attract individual investors, then we should expect advertising to have a greater impact in BtoC firms (where the advertising is targeted at individuals) rather than in BtoB firms (where the advertising is not targeted towards individuals). Second, investors are more likely to be aware of older firms that have been in existence for a long time, as compared to younger firms. Hence, we should expect advertising to
24
be more effective in reducing liquidity risk in the younger firms in P2 and P3. Third, if advertising reduces liquidity risk via its effect on individual investors, a mediation analysis would provide additional confirming evidence.
BtoB versus BtoC firms. Compared to a BtoC firm, if the advertising of a BtoB firm is targeted mainly towards trade journals and other media that have limited circulation among individuals, it will create less awareness among individual investors and we would expect it to be less likely to reduce liquidity risk. We create two size and advertising expenditure matched portfolios representing these two types of firms. Fama and French (1997) use four digit SIC codes to develop 48 industry categories that are used widely in finance research to form industry portfolios.vi We select firms that are in typically B2B (wholesale, machinery, lab equipment, construction material, and trading industries) or B2C (candy and soda, tobacco, consumer goods, and retail categories) industries. For each firm in the B2B category, we find a matching firm in the B2C category whose total assets and advertising expenses are within 70% to 130% of the B2B firm. We end up with an average of 104 matched firms per year in each of the B2B and B2C groups. We then divide the firms into two equal-sized groups based on their advertising expenditure. Thus, for each year between 1971 and 2004 we form two portfolios within the two industry groups. In both B2B and B2C industries, the average advertising for low- (high-) advertising portfolio firms is about $1.3 million ($25 million) and the total assets average about $80 million ($900 million). We expect that compared to B2B firms, the liquidity risk for B2C firms should decrease more as we move from the low-advertising to the high-advertising portfolio. Advertising by B2C firms should be visible to more investors as compared to advertising by B2B firms, since the
25
audience for advertising by B2C firms is typically broader. Over the 34-years estimation period (1972-2005), the difference between the high and low advertising portfolios is positive and usually statistically significant for annual (difference in liquidity risk; B2C firms, .341, p< .01; B2B firms, .357, p< .01) as well as pooled regressions (difference in liquidity risk; B2C firms, .187, p< .01; B2B firms, .028, ns). While the reduction in liquidity risk for BtoB and BtoC firms is similar when we use annual regressions, the reduction in liquidity risk is more for B2C firms than B2B firms using pooled regressions (difference in decrease in liquidity risk between B2B and B2C, .159, p< .05). These results offer some support for the notion that the visibility of the advertising to investors helps improve investor awareness and lowers liquidity risk.
Young versus old firms: We examine the impact of age by re-estimating the models in Table 4 separately for younger (age <= median age) and older (age > median age) firms in each of the four portfolios. The dependent variable is liquidity risk and the main explanatory variable of interest is advertising. We find that the coefficient on advertising is significantly negative only for the younger firms in P2 (-.154, p<.01) and in P3 (-.018, p<.05). The coefficient on advertising is negative but not significant at the usual levels for the older firms. As before, we continue to find that advertising does not lower liquidity risk for either young or old firms in P1 and P4; the coefficients on advertising in all these four models are not significantly different from zero. Thus, advertising appears to be effective in lowering liquidity risk only for younger firms, suggesting that improving awareness among individual investors and attracting them to relatively unknown firms is how advertising helps in reducing liquidity risk.
26
Mediation analysis: We conduct a mediation analysis to investigate whether the number of individual investors mediates the effect of advertising on liquidity risk. We test the conceptual model in Figure 1 using both the Baron and Kenny (1986) approach and the structural equations modeling approach (Zhao, Lynch and Chen 2010). We transform our panel data to meet the assumptions of these approaches, and confirm that the number of individual investors indeed fully mediates the link between advertising and LR. We also find that the number of institutional investors does not mediate liquidity risk. Taken together, these results strongly support H2 that liquidity risk will be lower for firms with more individual investors and that advertising likely increases the awareness of individual investors, and attracting them to invest in the firm
Economic Impact We have shown that advertising is negatively related to liquidity risk. However, increased advertising can create value in several ways, such as 1) increasing sales and thus cash flow, 2) creating brand loyalty, reducing cash flow variability and therefore decreasing systematic market risk, 3) increasing the liquidity level for a firm's stock and 4) increasing individual investor awareness and reducing liquidity risk. A reduction in market risk or liquidity risk and an increase in liquidity level will all lead to a lower cost of equity. Whether or not the incremental valuation impact of the reduction in liquidity risk is large will determine the importance of these results for practicing managers. We make some simplifying assumptions and estimate the partial impact of each effect on firm value one at a time (holding the other three effects constant). We estimate the impact of a 25% increase in advertising expenditure on market value for a typical firm in each of the four positive advertising portfolios.
27
To estimate the impact on sales, we assume that the advertising elasticity of demand is .2 (as in Leone 1995) for the average firm in our sample. Of course, we note that the effect on cash flow for an individual firm will be different and will depend on its advertising elasticity. The increased cash flows of the average firm in the portfolio are calculated as (operating margin*sales), net of advertising expenditure and taxes. We assume that the corporate tax rate is 35%. In order to estimate the impact on liquidity level we use estimates from Grullon, Kanatas and Weston (2004, average relative bid-ask spread calculated based on Table 4 panel A, advertising coefficient from Table 5 column 2). To assess the impact of increased advertising on reduction in market risk, we use the regression coefficients of market risk on advertising from McAlister, Srinivasan and Kim (2007, Table 2 column 1). The change in the firm's cost of equity capital in each portfolio due to a 25% increase in advertising is determined by multiplying the new market and liquidity risk by the average risk premium over our sample period. The firm value is estimated by dividing the cash flows by the estimated cost of equity (i.e., assuming perpetual cash flows). Insert Table 6 about here We first estimate the marginal impact of advertising on firm value via the cash flow effect, holding the cost of equity constant. In all four portfolios, advertising has a large impact on firm value through increased cash flows. This is expected since the first order impact of advertising on firm value is via increased sales. The cash flow impact on firm value varies between 4.63% (P1) and 6.62% (P3). Similarly, we estimate marginal changes in firm value due to reduction in cost of equity, assuming that there is no change in net cash flow. We note that this assumption implies that advertising increases sales to at least offset the increased cost of advertising. vii The impact on value due to reduction in market risk is much smaller and varies
28
between .09% (P1) and 1.10% (P4). Increases in liquidity levels are also associated with a small increase in firm value between .23% (P1) and .37% (P4). The increase in value due to change in liquidity risk varies between 0% (P1) to 1.39% (P3). In P1 there is no change in firm value due to reduction in liquidity risk. This result is not surprising when we consider that even a 25% increase in advertising is an economically small amount and is unlikely to increase individual investor awareness. In contrast, firms in the highest advertising quartile are already likely to be quite visible to the investors. Any increase in such firms' advertising expenditure is unlikely to increased investor awareness. Hence, for these firms also there should not be any marginal reduction in liquidity risk due to increased advertising. However, as discussed earlier, the small increase in liquidity risk leads to an estimated reduction of 1.27% in value. We note that these firms may still be creating incremental value by increasing advertising because the impact on value via increased cash flows is larger in magnitude (5.29%). The more interesting cases are the firms in P2 and P3, with medium levels of advertising. The increase in firm value due to reduction in liquidity risk is 1.30% in P2 and 1.39% in P3. This indicates that out of the four advertising portfolios, the firms in portfolio P3 gain the most due to the reduction in liquidity risk followed by the firms in portfolio P2. Based on P2's original valuation of $165.32 million an increase of 1.30% corresponds to an increase in market value of around $2.15 million. Similarly, from the base value of $455.91 million, the increase in P3's market value is around $6.34 million. Considering P2 and P3's incremental advertising expenditure of $ .35 million and $1.72 million respectively, the increase in value due to reduced liquidity risk is substantial. This suggests that the value relevance of advertising may have been underestimated in prior research.
29
Discussion and Conclusions
In this paper, we hypothesize and document that advertising can reduce liquidity risk through its impact on individual investor behavior. While Grullon, Kanastas and Weston (2004) examine the impact of advertising on liquidity level, our paper is the first to document its impact on liquidity risk. An economic impact analysis shows that while the cash flow impact of increasing advertising is large, reduction in liquidity risk also improves firm value. For a typical firm in quartiles P2 and P3, the improvement in firm value due to lower liquidity risk is about four to five times that due to increase in the liquidity level. Thus, it is important for marketers to understand not only the impact of advertising on cash flow, but also on liquidity risk. Our main contributions to the marketing literature are (a) to introduce liquidity risk as an important driver of firm value that can be affected by marketing actions such as advertising, (b) to show that the spillover effect of advertising (Lovett and MacDonald 2005, Joshi and Hanssens 2010) plays an important role in affecting individual investor awareness in capital markets, and (c) to suggest that methodologically, future researchers should include liquidity risk (which is a priced, systematic risk) in their models of expected returns. This reinforces the call of Tuli and Bharadwaj (2009) for empirical literature in marketing to focus on multiple dimensions of risk including systematic, idiosyncratic and downside risk. Managerial implications. Our results suggest that firms can increase value by attracting and retaining individual investors. Firms typically communicate with the investor community through the required disclosures of financial information (e.g., financial statements, 10K reports, etc.) as well as by establishing an investor relations (IR) strategy with the ?goal of creating an understanding of the firm, attracting information intermediaries, and targeting a desired investor
30
base? (Brennan and Tamarowski 2000). The creation of the IR department results in increased firm visibility through higher media coverage and analyst following as well as more institutional investors (Bushee and Miller 2007). While the IR activities mainly focus on the institutional investor community, our findings suggest that IR departments should also pay attention to communicating with individual investors. Marketing managers make decisions both on the level of advertising expenditures as well as on the media-mix that determines the allocation of this money across various media. It appears that advertising is effective in reducing liquidity risk for firms (in P2 and P3) that have annual advertising budget between about $ 0.5 million and $20 million. Firms spending outside this range will not see much reduction in their liquidity risk due to increased ad-spending. This impact may be especially important for younger firms, as well as BtoC firms as their advertising can be more effective in reducing liquidity risk. Even if the advertising budgets cannot be increased, our results imply that firms can carefully select the advertising media mix to achieve their goal of targeting individual investors. Specifically, if firms can redirect some of the product advertising to media that simultaneously targets individuals who are potential investors, this could also increase awareness among individual investors. Firms may advertise their products directly in the financial press (e.g., Wall Street Journal) or in media that are commonly perused by individual investors (e.g., CNBC, local print media, etc.). For example, Kodak advertised their ink saving inkjet printers and Sony advertised its e-book reader using full page advertisements in the Wall Street Journal (March 30, 2009). Firms may also be able to develop specific advertising and marketing campaigns aimed at individual investors and encourage them to invest in the firm. This may be even more important in times of market stress. In addition to media mix, the timing of the advertising may also impact liquidity risk. Media planners use
31
pulsing of advertising expenditures to increase their effectiveness (Leckenby and Kim 1994). But this can result in cash flow volatility, leading to higher risk (Fischer, Shin and Hanssens 2009). Analogously, pulsing might also increase the fluctuations in awareness among individual investors, which could lead to increased liquidity risk. Managers may need to take this additional negative financial impact into consideration as they evaluate the timing of their marketing spending policies. Our results also highlight a crucial difference between liquidity level and liquidity risk. Specifically, if a firm is only interested in increasing the liquidity level of its stock, then ownership structure may be less important and it may not matter whether the higher liquidity level comes about due to increased trading by institutional or individual investors. However our results indicate that a larger pool of institutional (individual) investors increases (decreases) liquidity risk, which may reduce (augment) any benefit the firm may obtain by increasing liquidity level. Since the impact of reducing liquidity risk on value is substantial compared to that of increasing liquidity level, our results suggest that firms should focus more on increasing the number of individual investors. This would both increase the liquidity level and decrease liquidity risk. Our findings, coupled with other findings about impact of advertising on systematic risk (McAlister, Srinivasan and Kim 2007), should also make it easier for marketing executives to justify their advertising expenditures to the financial executives, who are typically skeptical about the financial returns to these investments. We believe that our findings about the relation between liquidity risk and investor type are also new to the finance literature. By documenting that advertising can help lower the cost of equity capital by lowering both market and liquidity risks, the marketing department can more accurately quantify the full value of advertising and
32
related marketing expenditures. Managers may be able to reduce the cost of equity by keeping the advertising budget the same but altering the media mix. The lower cost of capital would improve the viability of hitherto unprofitable marketing projects such as brand extensions, new product introductions, etc. Limitations and future research. The effectiveness of the firm in raising individual investor awareness through product market advertising will depend on the branding strategy of the firm. Advertising should be more effective for firms that follow a corporate branding strategy compared to a house-of-brands strategy, as investors can link the brands and products of the corporate brand with the firm more easily. It was not possible for us to code the branding strategy for such a large number of firms, and we leave this to future research. The efficacy of different types of media, e.g., TV, print, internet, etc., is also likely to be different in raising investor awareness. A number of other marketing strategies, besides advertising, can also impact investor awareness. For example, a firm targeting a broader market segment or with an intensive distribution is likely to be more visible to investors. Similarly, the CSR actions of a firm can increase the likelihood that socially responsible individual investors would be willing to invest in the firm. The capital market impact of these and other marketing actions need to be studied more carefully to fully understand the impact of marketing actions on liquidity risk and firm value.
33
References Acharya, Viral V. and Lasse Heje Pedersen (2005), ?Asset Pricing with Liquidity Risk,? Journal of Financial Economics, 77 (2), 375-410.
Amihud, Yakov (2002), ?Illiquidity and Stock Returns: Cross-Section and Time-Series Effects,? Journal of Financial Markets, 5 (1), 31-56.
—— and Haim Mendelson (1986), ?Asset Pricing and the Bid-Ask Spread,? Journal of Financial Economics, 17 (2), 223-49.
Barber, Brad M. and John D. Lyon (1997), ?Detecting Long-Run Abnormal Stock Returns: The Empirical Power and Specification of Test Statistics,? Journal of Financial Economics, 43 (3), 341-72.
Baron, R. M. and D. A. Kenny (1986), ?The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations,? Journal of Personality and Social Psychology, 51 (6), 1173-82.
Beaver, William, Paul Kettler, and Scholes Myron (1970), ?The Association between Market Determined and Accounting Determined Risk Measures,? The Accounting Review, 45 (4), 65482.
34
——, Maureen McNichols, and Richard Price (2007), ?Delisting Returns and Their Effect on Accounting-Based Market Anomalies,? Journal of Accounting and Economics, 43 (2-3), 341-68. Brennan, Michael J. and Claudia Tamarowski (2000), ?Investor Relations, Liquidity, and Stock Prices,? Journal of Applied Corporate Finance, 12 (4), 26-37.
Bushee, Brian J. and Gregory S. Miller (2007), ?Investor Relations, Firm Visibility, and Investor Following,? SSRN eLibrary Working Paper.
Carhart, Mark M. (1997), ?On Persistence in Mutual Fund Performance,? Journal of Finance, 52 (1), 57-82.
Chordia, Tarun, Richard Roll, and Avanidhar Subrahmanyam (2000), ?Commonality in Liquidity,? Journal of Financial Economics, 56 (1), 3-28.
Dennis, Patrick J. and Strickland Deon (2002), ?Who Blinks in Volatile Markets, Individuals or Institutions?,? The Journal of Finance, 57 (5), 1923-49.
Eckbo, B. Espen and Øyvind Norli (2005), ?Liquidity Risk, Leverage and Long-Run Ipo Returns,? Journal of Corporate Finance, 11 (1-2), 1-35.
Fama, Eugene F and Kenneth R. French (1993), ?Common Risk Factors in the Returns on Stocks and Bonds,? Journal of Financial Economics, 33 (1), 3-56.
35
—— and —— (1997), ?Industry Costs of Equity,? Journal of Financial Economics, 43 (2), 15393.
—— and —— (2004), ?The Capital Asset Pricing Model: Theory and Evidence,? Journal of Economic Perspectives, 18 (3), 25-46.
—— and James D. MacBeth (1973), ?Risk, Return, and Equilibrium: Empirical Tests,? Journal of Political Economy, 81 (3), 607-36.
Fehle, F., S. Tsyplakov, and V. Zdorovtsov (2005), ?Can Companies Influence Investor Behaviour through Advertising? Super Bowl Commercials and Stock Returns,? European Financial Management, 11 (5), 625-47.
Fischer, Marc, Hyun Shin, and Dominique M. Hanssens (2009), ?Marketing Spending and the Volatility of Revenues and Cash Flows,? Working Paper (UCLA Anderson School of Management).
Fornell, Claes, Sunil Mithas, Forrest V. Morgeson III, and M. S. Krishnan (2006), ?Customer Satisfaction and Stock Prices: High Returns, Low Risk,? Journal of Marketing, 70 (1), 3-14.
Frieder, Laura and Avanidhar Subrahmanyam (2005), ?Brand Perceptions and the Market for Common Stock,? Journal of Financial and Quantitative Analysis, 40(1), 57-85.
36
Gruca, Thomas S. and Lopo L. Rego (2005), ?Customer Satisfaction, Cash Flow, and Shareholder Value,? Journal of Marketing, 69 (3), 115-30.
Grullon, Gustavo, George Kanatas, and James P. Weston (2004), ?Advertising, Breadth of Ownership, and Liquidity,? Review of Financial Studies, 17 (2), 439-61.
Hanka, Gordon (1998), ?Debt and the Terms of Employment,? Journal of Financial Economics, 48 (3), 245-82.
Huberman, Gur (2001), ?Familiarity Breeds Investment,? Review of Financial Studies, 14 (3), 659-80.
Joshi, Amit and Dominique M. Hanssens (2009), ?The Direct and Indirect Effects of Advertising Spending on Firm Value,? Journal of Marketing, 74 (1), 20-33.
Kaniel, Ron, Gideon Saar, and Sheridan Titman (2008), ?Individual Investor Trading and Stock Returns,? The Journal of Finance, 63 (1), 273-310.
Korajczyk, Robert A. and Ronnie Sadka (2008), ?Pricing the Commonality across Alternative Measures of Liquidity,? Journal of Financial Economics, 87 (1), 45-72.
Leckenby, John D. and Heejin Kim (1994), ?How Media Directors View Reach/Frequency Estimation: Now and a Decade Ago,? Journal of Advertising Research, 34 (5), 9-21.
37
Lee, Kuan-Hui (2010), ?The World Price of Liquidity Risk,? Journal of Financial Economics, Forthcoming.
Leone, Robert P. (1995), ?Generalizing What Is Known About Temporal Aggregation and Advertising Carryover,? Marketing Science, 14 (3), G141-G50.
Lin, Ji-Chai, Ajai K. Singh, and Wen Yu (2009), ?Stock Splits and the Trading Speed Improvement Hypothesis,? Journal of Financial Economics, 93 (3), 474-89.
Lintner, John (1965), ?The Valuation of Risk Assets and the Selection of Risky Investments in Stock Portfolios and Capital Budgets,? The Review of Economics and Statistics, 47 (1), 13-37.
Litzenberger, Robert H. and Krishna Ramaswamy (1979), ?The Effects of Dividends on Common Stock Prices: Theory and Empirical Evidence,? Journal of Financial Economics, 7 (2), 163-95.
Liu, Weimin (2006), ?A Liquidity-Augmented Capital Asset Pricing Model,? Journal of Financial Economics, 82 (3), 631-71.
Lovett, Mitchell J. and Jason B. MacDonald (2005), ?How Does Financial Performance Affect Marketing? Studying the Marketing-Finance Relationship from a Dynamic Perspective,? Journal of the Academy of Marketing Science, 33 (4), 476-85.
38
Lundgren, Terry J. (2007), ?Federated Plans Corporate Name Change,? (accessed April 22, 2009), [available at http://phx.corporate-ir.net/phoenix.zhtml?c=84477&p=irolnewsArticle&ID=967632&highlight=].
Luo, Xueming (2007), ?Consumer Negative Voice and Firm-Idiosyncratic Stock Returns,? Journal of Marketing, 71 (3), 75-88.
——and CB Bhattacharya (2009), ?The Debate over Doing Good: Corporate Social Performance, Strategic Marketing Levers, and Firm-Idiosyncratic Risk,? Journal of Marketing, 73 (6), 198-213.
McAlister, Leigh, Raji Srinivasan, and Min Chung Kim (2007), ?Advertising, Research and Development, and Systematic Risk of the Firm,? Journal of Marketing, 71 (1), 35-48.
Merton, Robert C. (1987), ?A Simple Model of Capital Market Equilibrium with Incomplete Information,? Journal of Finance, 42 (3), 483-510.
Mitra, Anusree and John G. Lynch, Jr. (1995), ?Toward a Reconciliation of Market Power and Information Theories of Advertising Effects on Price Elasticity,? Journal of Consumer Research, 21 (4), 644-59.
39
Mizik, Natalie and Robert Jacobson (2007), ?Myopic Marketing Management: Evidence of the Phenomenon and Its Long-Term Performance Consequences in the Seo Context,? Marketing Science, 26 (3), 361-79.
Pástor, Luboš and Robert F. Stambaugh (2003), ?Liquidity Risk and Expected Stock Returns,? Journal of Political economy, 111 (3), 642-85.
Sadka, Ronnie (2006), ?Momentum and Post-Earnings-Announcement Drift Anomalies: The Role of Liquidity Risk,? Journal of Financial Economics, 80 (2), 309-49.
Schwartz, Jonathan (2007), ?The Rise of JAVA - the Retirement of SUNW,? (accessed April 22, 2009), [available at http://blogs.sun.com/jonathan/entry/java_is_everywhere].
Sharpe, William F. (1964), ?Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk,? The Journal of Finance, 19 (3), 425-42.
Shumway, Tyler (1997), ?The Delisting Bias in Crsp Data,? Journal of Finance, 52 (1), 327-40.
—— and Vincent A. Warther (1999), ?The Delisting Bias in Crsp's Nasdaq Data and Its Implications for the Size Effect,? The Journal of Finance, 54 (6), 2361-79.
40
Sorescu, Alina, Venkatesh Shankar, and Tarun Kushwaha (2007), ?New Product Preannouncements and Shareholder Value: Don't Make Promises You Can't Keep,? Journal of Marketing Research, 44 (3), 468-89.
—— and Jelena Spanjol (2008), ?Innovation's Effect on Firm Value and Risk: Insights from Consumer Packaged Goods,? Journal of Marketing, 72 (2), 114-32.
Srinivasan, Shuba and Dominique M. Hanssens (2009), ?Marketing and Firm Value: Metrics, Methods, Findings, and Future Directions,? Journal of Marketing Research, 46 (3), 293-312.
Tuli, Kapil R. and Sundar G. Bharadwaj (2009), ?Customer Satisfaction and Stock Returns Risk,? Journal of Marketing, 73 (6), 184-97.
——, Sundar G. Bharadwaj, and Ajay K. Kohli (2010), ?Ties That Bind: The Impact of Multiple Types of Ties with a Customer on Sales Growth and Sales Volatility,? Journal of Marketing Research, 47 (1), 36-50.
Vakratsas, Demetrios and Tim Ambler (1999), ?How Advertising Works: What Do We Really Know?,? Journal of Marketing, 63 (1), 26-43.
Zhao, Xinshu, John G. Lynch, Jr., and Qimei Chen ?Reconsidering Baron and Kenny: Myths and Truths About Mediation Analysis,? Journal of Consumer Research, 37 (August), 197-206.
41
TABLE 1 Descriptive Statistics of the Sample Firms (1971 through 2005) MVE (Mill. $) Total Assets (Mill. $) Sales Advertising Advertising /Sales R&D /Sales Leverage Firm Age (Years) 11 .84 10 .89 11 .79 Q3 13 .08 N 1705.69 6.75 5.72 6.26Q3 7.52N 106.57 6.11 4.88 6.14Q3 7.02N 426.03 7.84 7.13 7.74Q3 8.38N 426.49 10 .73 9.48 10 .72 12 .09 N 426.89 22 .67 21 .30 22 .12 24 .86 N 426.29
Mean Q1 Median
Mean Q1 Median
Mean Q1 Median
Mean Q1 Median
Mean Q1 Median Q3
Mean Q1 Median Q3
(Mill. $) (Mill. $) A: Firms with Positive Advertising 1592.12 2253.57 1551.89 45 .36 0.07 0.58 710.80 1069.43 1306.83 36 .75 0.03 0.05 1070.94 1699.07 1425.97 42 .32 0.04 0.23 2308.84 2298.88 1806.41 57 .97 0.07 0.57 1705.69 1705.69 1705.69 1705.69 1705.69 740.80 B : Firms with Zero Advertising 295.14 1120.18 126.49 0.00 0.00 2.48 67 .09 92 .33 41 .39 0.00 0.00 0.05 139.31 211.37 85 .57 0.00 0.00 0.61 331.14 1674.37 228.70 0.00 0.00 3.26 106.57 106.57 106.57 106.57 106.57 7.54 C: Firms in the First Advertising Quartile (Lowest Advertising) - Portfolio P1 70 .15 115.98 61 .70 0.25 0.07 1.45 34 .27 26 .09 19 .89 0.15 0.01 0.11 51 .94 129.70 38 .93 0.21 0.04 0.53 77 .17 178.92 87 .73 0.28 0.06 1.47 426.03 426.03 426.03 426.03 426.03 206.86 D: Firms in the Second Advertising Quartile - Portfolio P2 165.32 295.44 148.48 1.38 0.09 0.59 75 .59 96 .30 85 .37 0.92 0.02 0.03 108.87 240.68 114.41 1.17 0.04 0.16 240.68 511.34 188.85 1.60 0.07 0.39 426.49 426.49 426.49 426.49 426.49 198.37 E: Firms in the Third Advertising Quartile - Portfolio P3 455.91 773.92 464.94 6.88 0.05 0.13 235.79 388.73 272.66 5.23 0.03 0.02 286.79 528.35 436.77 6.47 0.04 0.07 641.03 960.34 614.96 7.45 0.05 0.16 426.89 426.89 426.89 426.89 426.89 163.03 F: Firms in the Fourth Advertising Quartile (Highest Advertising) - Portfolio P4 5679.30 7831.36 5534.03 172.99 0.07 0.05 2410.23 3621.39 4574.24 137.71 0.05 0.03 3962.36 5750.04 4990.25 163.29 0.05 0.04 7924.37 7684.04 6611.20 221.06 0.06 0.06 426.29 426.29 426.29 426.29 426.29 172.54
0.30 0.25 0.28 0.33 1705.54 0.30 0.22 0.30 0.38 106.51 0.26 0.20 0.24 0.30 425.91 0.30 0.25 0.29 0.33 426.46 0.32 0.27 0.30 0.38 426.89 0.30 0.26 0.29 0.33 426.29
Notes: Each year we compute sample average of each variable. In the above table we report the time series mean, quartile1 (Q1), median, quartile3 (Q3), and sample size (N) of each variable average computed over 35 years period. All the dollar amounts are adjusted to 2005 US dollars using Consumer Price Index (CPI; obtained from the Website of Federal Reserve Bank of St. Louis [http://research.stlouisfed.org/]). MVE is market value of equity.
42
TABLE 2 Estimates of Liquidity Risk for Zero and Positive Advertising Portfolios (One-Year Advertising Portfolio - One-Year Estimation Period) Advertising Portfolios Zero Advertising Positive Advertising (Z1) Model 1 Market Risk Liquidity Risk Adj. R2 Model 2 Market Risk Liquidity Risk SMB HML UMD Adj. R2
a
Difference in Risks (Z1) - (Z2) .075 (.65 ) .4 7 9a (2.89)
(Z2) 1 .2 9 6a (18.45) .4 3 4a (4.74) .7 0
1 .3 7 1a (10.06) .9 1 3a (5.47) .4 0
1 .3 2 4a (8.99) .9 6 3a (4.18) .8 0 3a (4.27) -.041 (-.21) -.215 (-1.09) .6 4
1 .1 0 6a (26.25) .3 8 0a (6.18) .9 9 9a (18.70) -.085 (-1.15) -.233a (-4.95) .9 5
.217 (1.57) .5 8 3b (2.46) -.195 (-1.05) .044 (.22 ) .017 (.09 )
denotes p<= .01 denotes p<= .05 denotes p<= .1 (two-sided) t-values are in parentheses, year dummy coefficients not reported Model 1 is liquidity risk augmented CAPM which comprises market factor and Liu's liquidity factor Model 2 is Carhart 4-factor model which comprises market factor (Market), size factor (SMB), book-to-market factor (HML), momentum factor (UMD) and liquidity factor (Liquidity)
(two-sided), b
(two-sided), c
43
TABLE 3 Estimates of Liquidity Risk for Positive Advertising Portfolios (OneYear Advertising Portfolio - One-Year Estimation Period) Advertising Portfolios Lowest Advertising (P1) Model 1 Market Risk Liquidity Risk Adj. R2 Model 2 Market Risk Liquidity Risk SMB HML UMD Adj. R2
a
Highest Advertising (P2) 1 .3 2 4a (14.47) .4 8 9a (4.25) .6 2 (P3) 1 .2 4 5a (17.21) .2 6 6b (2.55) .7 1 (P4) 1 .1 6 1a (26.59) .1 1 2c (1.81) .8 1 Difference in Risks (P1-P2) .1 3 2c (2.02) .3 8 0a (4.81) (P1-P3) .2 1 0a (3.28) .6 0 3a (6.77) (P1-P4) .2 9 5a (3.33) .7 5 7a (6.57)
1 .4 5 6a (14.11) .8 6 9a (6.80) .5 5
1 .1 9 8a (14.85) .9 1 9a (9.59) 1 .1 6 4a (13.90) -.209c (-1.73) -.296a (-3.59) .8 5
1 .0 7 2a (23.29) .4 0 3a (4.62) 1 .1 8 9a (16.39) -.160c (-1.73) -.238a (-3.47) .9 1
1 .0 9 3a (21.14) .2 0 7a (2.73) 1 .0 4 8a (16.27) -.059 (-.79) -.235a (-5.00) .9 4
1 .0 6 2a (32.16) -.010 (-.20) .5 9 3a (11.49) .088 (1.46) -.162a (-3.40) .9 4
.1 2 6c (1.93) .5 1 7a (5.86) -.025 (-.38) -.049 (-.62) -.057 (-.92)
.105 (1.49) .7 1 3a (7.19) .116 (1.27) -.150 (-1.56) -.061 (-.80)
.1 3 6c (1.72) .9 3 0a (10.30) .5 7 1a (7.32) -.297b (-2.57) -.134 (-1.41)
denotes p<= .01 (two-sided), b denotes p<= .05 (two-sided), c denotes p<= .1 (two-sided) t-values are in parentheses, year dummy coefficients not reported Model 1 is liquidity risk augmented CAPM which comprises market factor and Liu's liquidity factor Model 2 is Carhart 4-factor model which comprises market factor, liquidity factor, size factor (SMB), book-to-market factor (HML), and momentum factor (UMD),
44
TABLE 4 Impact of Advertising on Liquidity Risk at the Firm-Level (One-Year Advertising Portfolio - Three-Year Estimation Period) (1971-2005) P1 Intercept Advertising Total Assets Leverage Book to Market Age R2 N Unique Firms
a
P2
a
P3 -.202 (-3.32) (-2.54) -.000 (-.70) .3 3 7a (3.30) .004 (.15 ) -.011 (-1.10) .1 0 8490 1890
a
P4 -.126a (-2.65) .0 0 0bx (2.38) .000 (.45 ) .2 2 4a (2.76) .1 0 7a (5.19) .0 2 9a (4.81) .0 6 10743 1324
.493 (4.68) .134 (.77 ) -.000 (-.28) .219 (1.41)
.026 (.33 ) -.078b (-2.11) -.000c (-1.78) .070 (.53 ) .0 8 9a (2.78) .012 (.79 ) .0 7 7106 2019
-.014b
.0 6 6c (1.73) .009 (.37 ) .0 6 6001 1895
denotes p<= .01 (two-sided), b denotes p<= .05 (two-sided), c denotes p<= .1 (two-sided) xActual coefficient is 0.00047 t-values are in parentheses, year dummy coefficients not reported Advertising and total assets are adjusted to 2005 US dollars using Consumer Price Index
45
Table 5 Impact of Number of Investors on Liquidity Risk at the Firm Level Liquidity Risk .072 (.81 ) Log (No. of Institutional Investors) Log (No. of Individual Investors) Total Assets Leverage Book to Market Age Profit Margin (%) Market Share (%) B 2B B 2C .0 3 9c (1.90) -.021c (-1.80) -.000a (-7.90) .054 (.77 ) .0 3 6b (2.21) -.004 (-.32) -.000a (-2.82) .010(.5 5 )-.045 (-.69) .143 (1.03) .0 3 23013 5212
Intercept
R2 N Unique Firms
a
denotes p <= .01 (two-sided), b denotes p <= .05 (two-sided), c denotes p <= .1 (two-sided) Year dummy coefficients not reported.
46
TABLE 6 Economic Effect of Increase in Advertising on Firm Value via Impact on Cash Flow, Market Risk, Liquidity Level and Liquidity Risk P1 70 .15 61 .70 .2 5 38.10% 35.00% .49% .51% .72% 1 .456 .869 22.25% 15 .61 .0 6 3.09 .000 .869 22.25% .00% .7 2 4.63% -3.187 1 .453 22.23% .09% -.047 -1.18% 4.33% -.05% 22.20% .23% P2 165.32 148.48 1.38 40.50% 35.00% .49% .51% .72% 1 .324 .489 18.15% 30 .00 .3 5 7.42 -.078 .462 17.91% 1.30% 1.73 5.77% -3.187 1 .317 18.10% .25% -.047 -1.18% 4.33% -.05% 18.10% .28% P3 455.91 464.94 6.88 38.84% 35.00% .49% .51% .72% 1 .245 .266 15.73% 71 .72 1.72 23 .25 -.014 .241 15.51% 1.39% 4.75 6.62% -3.187 1 .233 15.66% .46% -.047 -1.18% 4.33% -.05% 15.68% .32% P4 5679.30 5534.03 172.99 38.81% 35.00% .49% .51% .72% 1 .16110 .11211 13.88% 788.50 43 .25 276.70 .00016 .13317 14.06% -1.27% 41 .69 5.29% -3.187 1 .13623 13.73% 1.10% -.047 -1.18% 4.33% -.05% 13.83% .37%
1 2 3 4 5 6 7 8 9
Market Value of Equity (MVE; M$) (Our Table 1) Sales (M$) (Our Table 1) Advertising (M$) (Our Table 1) Gross Margina,b Tax rate Monthly Risk-free Ratea,c Monthly Market Risk Premiuma,ca,d Monthly Liquidity Risk Premium Market Beta (Our Table 3) Liquidity Beta (Our Table 3) Annual Cost of Equity = 12 x {(6) + ((7) x (9)) + ((8) x (10))} Estimated Cash Flows (M$) = (1) x (11) Increase in Advertising (M$) = 25% x (3) Increase in Sales (M$) = .2 x 25% x (2) ? Liquidity Beta (Our Table 4) Liquidity Beta After 25% Incr. in Advt. = (10) + ((13) x (15)) New Cost of Equity = 12 x {(6) + ((7) x (9)) + ((8) x (16))} % ? MVE Due to Liquidity Risk = [(12)/(17) - (1)]/(1) ? Post-Tax Cash flow = {1 - (5)} x {(4) x (14)} % ? MVE Due to Cash Flow = [(19)/(11)] / (1) McAlister et al. Coefficient (Their Table 2 Column 1) New Market Beta = (9) + {(21) x ((13) / (2))} New Cost of Equity = 12 x [(6) + ((7) x (22)) + ((8) x (10))] % ? MVE Due to Mkt. Risk = [(12)/(23) - (1)] / (1) Coefficient of Log Advertising on Log Relative Spread from Grullon et al. (Their Table 5, Column 2) % ? Rel. Spread Due to Increase in Advt. = 25% x (25) Average eRelative Spread for Grullon et. al. (Their Table 4, Panel A) Decrease in Relative Bid-Ask Spread = (26) x (27) New Annual Cost of Equity = (11) + (28) % ? MVE Due to Liquidity Level = [(12) / (29) - (1)] / (1)
12 13 14 15
18 19 20 21 22 24 25 26 27 28 29 30
a b
Average for time period 1971-2005 Computed as (Sales - Cost of Goods Sold - Advertising)/Sales c Monthly values of risk free rate and market risk premium as reported on WRDS d Monthly values of liquidity risk premium from Liu e Average of relatinve bid ask spreads of the 25 portfolios in Panel A
47
Figure 1 - Conceptual model
Product Market Outcomes Market Share and Profit Margin Advertising Expenditure Investor Awareness
Investor Pool No. of Institutional Investors Liquidity Risk No. of Individual Investors
48
FIGURE 2 PORTFOLIO FORMATION AND ESTIMATION
Portfolio formation period = 1 year, Estimation period = 1 year
1972 1973 1974
?? ? ?? ?? ? ?? ? ? ?? ?
34 Estimation Periods
2005
Portfolio formation period = 3 years, Estimation period = 1 year
1974 1975 1976
?? ? ?? ?? ? ?? ? ? ?? ?
32 Estimation Periods
2005
1971
1972 1973 1974 1975 1976 1977
--------------
2001 2002 2003 2004 2005
Portfolio formation period
Estimation period
49
FIGURE 3 DIFFERENCE IN LIQUIDITY RISKS OF ADVERTISING PORTFOLIOS (ONE-YEAR ADVERTISING PORTFOLIO - ONE-YEAR ESTIMATION PERIOD) Liquidity Risk of Zero Advertising Portfolio (Z1) - Liquidity Risk of Positive Advertising Portfolio (Z2)
50
FIGURE 4 DIFFERENCE IN LIQUIDITY RISKS OF ADVERTISING PORTFOLIOS (ONE-YEAR ADVERTISING PORTFOLIO - ONE-YEAR ESTIMATION PERIOD) Liquidity Risk of Portfolio 1 (P1) - Liquidity Risk of Portfolio 2 (P2)
Liquidity Risk of Portfolio 1 (P1) - Liquidity Risk of Portfolio 3 (P3)
Liquidity Risk of Portfolio 1 (P1) - Liquidity Risk of Portfolio 4 (P4)
51
FIGURE 5 DECILE PORTFOLIO LIQUIDITY RISK AND MEDIAN ADVERTISING EXPENDITURES
5
10
15
20
25
30
35
40
383
52
Web Appendix
Liu's Liquidity Risk Factor (LIQ) Liquidity refers to the ease of trading a stock at the market price, at low cost, in sufficient quantities quickly (Liu 2006). The ability to trade a stock at the market price is also called the price impact of a trade. Generally, a trade in a liquid stock does not impact its market price, while an illiquid stock's trade can change the price a lot, making it less likely that an investor can trade his full desired quantity at the market price. In case the price impact is very high, an investor may decide to not trade at all. The bid-ask spread of a stock represents the cost of the trade, with the illiquid stocks having a higher bid-ask spread. Finally, an investor may be unable to find a counter party to trade an illiquid stock. Researchers use various types of data to capture the liquidity of a stock. Most are based on actual trading data - e.g. turnover, bid-ask spread etc. If the measures do not account for the number of zero-trading days, then the measure does not comprehensively capture the notion of ease of trading. Since the liquidity risk measures are derived from the underlying liquidity level measures, the corresponding liquidity risk measure is also likely to be less complete. Liu (2006) first calculates the individual stock liquidity factor LM12, and then uses LM12 to develop the market liquidity factor LIQ.
Liu's liquidity measure (LM12): LM12 is computed for each stock based on the data from the previous 12 months. LM12 Number of zero trading days in prior 12 months 1/(12 month turnover ) * 21*12 Deflator TD
53
The 12-month turnover is the stock's turnover calculated as the sum of daily turnover over the prior 12 months. Daily turnover is the ratio of number of shares traded on a day to the number of shares outstanding at the end of the day. Twenty-one represents the average number of trading days in a month. TD is the total number of the trading days in the market over the prior 12 months. Deflator is a number chosen such that 0
1/(12 month turnover ) Deflator 1, for all sample
stocks. Liu chooses a deflator value of 11,000 for constructing LM12.
Liu's liquidity risk factor (LIQ): The market-wide liquidity factor, LIQ, is constructed using mimicking portfolio approach similar to that of Fama and French (1993). Each year, the stocks are sorted on their LM12 value and grouped into three equal-weighted portfolios. LIQ is computed as the difference between the monthly stock returns of the most illiquid and most liquid portfolios. The values of LIQ are available on a monthly basis.
Comparison with other liquidity factors: Several other liquidity measures have been proposed in the finance literature (Amihud 2002, Pástor and Stambaugh 2003, Sadka 2006). All these measures are based on actual trading data, and hence do not fully account for zero trading days. For example, Amihud (2002) estimates the ratio of daily absolute return to daily dollar trading volume averaged over a 12 month period. If there is no trading, then this ratio cannot be calculated. Pástor and Stambaugh (2003) derive their measure based on daily trading data over a month with a minimum of 16 trading days. If a stock does not trade or trades on less than 16 days in a month, then its liquidity measure cannot be estimated. The LM12 liquidity measure is a
composite measure that captures the inability of the investor to trade when needed. We believe that LM12 is a more appropriate liquidity measure than measures such as the Pastor-Stambaugh or the Sadka
54
measure. The latter two measures primarily capture the price impact of trading, which is only one dimension of liquidity. Thus we feel that Liu's (2006) measure captures more fully the non-trading
risk to an investor.
Accounting for Delisted Firms Many firms delist from stock exchanges each year for various reasons. Shumway(1997) and Shumway and Warther (1999) report that on average 1.2% NYSE and AMEX firms and 5.6% NASDAQ firms delist each year for performance related reasons (see Beaver, McNichols, and Price 2007 for a review). Once a firm delists, it stops trading on the stock exchange and CRSP sets the firm's subsequent stock returns to missing values. If delisting is not accounted for in the portfolio returns calculation, it can introduce a bias in the portfolio returns. For delisted firms, CRSP reports the delisting code and, if available, delisting returns. Delisting code provides the reason for delisting. Delisting return is the return from the firm's delisting date to its first off-exchange trade in the OTC (over the counter) markets. CRSP does its own research to estimate delisting returns. When the delisting return is unavailable, CRSP sets the delisting return to missing. Shumway (1997) shows that ignoring missing delisting returns biases the portfolio returns. Consider a portfolio consisting of two NYSE firms, A and B. Suppose that firm B delists in month t and trades subsequently in OTC markets, and has an actual delisting return of -30%. Example: Actual (equally-weighted) portfolio return
Price Firm A Firm B Portfolio of A and B Month t 100 100 200 Month t+1 Return 110 70 180 10% -30% -10% Month t 100 100 200 Month t+1 110 0 110 Return 10% -100% -45%
55
The actual portfolio return is -10%. But, if the return for firm B is missing on CRSP and is mistakenly coded as -100% by the researcher, this introduces a downward bias in the return to -45%. If the researcher omits this firm from analysis in month t+1, then the portfolio return is biased upwards to 10%, since only stock A is part of the portfolio. Shumway (1997) and Shumway and Warther (1999) obtained the average delisting returns by examining the prices of the firms on OTC (over the counter) markets. We follow Shumway (1997) and Shumway and Warther (1999) to adjust the missing delisting returns. We proceed as follows. If delisting return is not available and reason for delisting is performance related (delisting code between 500 and 599), we follow Shumway (1997) and Shumway and Warther (1999) and set the missing delisting returns to -30% for NYSE and AMEX firms and to -55% for NASDAQ firms in the delisting month. The firm is not considered as part of the portfolio in subsequent months.
56
FOOTNOTES
i
Portfolio theory (e.g., Lintner 1965; Sharpe 1964) posits that investors can diversify away firm-specific risks by investing in a portfolio of stocks. In equilibrium, the risks that are priced in the stock market will be systematic risks. We estimate liquidity risk as the slope coefficient in a regression of returns on a liquidity factor, similar to how the market beta is estimated as the slope coefficient in a regression of returns on the market factor. The exact estimation details are discussed later in the paper.
ii
It is possible that when a buyer or seller has valuable information and wants to trade quickly before other investors (i.e., they desire to trade immediately and demand liquidity), there may be no one to take the opposite side (i.e., supply liquidity). In practice, liquidity can be provided either by investors who want to trade but do not have a need to trade immediately (e.g., because they do not have short lived information) or by professional traders such as market makers on the NYSE and NASDAQ.
iii
Because the number of market makers is not affected in the short term by marketing activities, we do not focus on them. Also, as discussed earlier, rather than providing liquidity, trading by institutional investors tends to use up liquidity. Hence, attracting institutional investors is not likely to reduce liquidity risk.
iv
Several other liquidity risk factors have been proposed in the literature (e.g. Amihud 2002, Pástor and Stambaugh 2003, Sadka 2006). However the liquidity measures used to develop these risk factors use information from actual trades and do not include any data which measures how often there is no trading at all. Thus their liquidity risk measures do not fully capture an investors' difficulty of trading.
v
We also estimate the relationship between advertising and the number of individual and institutional investors for the period 1997-2005, using fixed effect panel model with AR(1) serial correlation in errors. We control for product market impacts (market share, profitability, BtoB and BtoC) as well as other variables suggested in Grullon, Kanatas and Weston (2004). We find that advertising positively impacts only the number of individual investors. Results are not separately tabulated, but are available upon request.
vi
Whenever available, we use the historical SIC code from CRSP to classify firms into the Fama French industries. If missing, we use the current SIC code as reported on COMPUSTAT (Hanka 1998).
vii
We have derived our estimates assuming cash flow neutrality, i.e., the incremental advertising cost is offset by increases in sales. We caution that if a firm desires to increase advertising purely to decrease cost of capital without any change in sales, this may not be an economically viable decision. Individual firm impacts will depend on their advertising sales elasticity.
57
doc_724633196.docx