1. Introduction

Research on the marketing-finance interface has had a significant development since 2004 (^{Edeling, Srinivasan, & Hanssens, 2021}). It has also evidenced a need to demonstrate value creation through advertising actions due to the growth of investments in this field and the measured and produced effects on firm performance and firm value (^{Joshi & Hanssens, 2010}; ^{Bayer, Srinivasan, Riedl, & Skiera, 2020}).

The market microstructure theory (^{Hasbrouck, 2007}) predicts that asymmetric information is related to stock returns, thus stimulating the development of several valid measures for it (^{Roll, 1984}; ^{Easley, Hvidkjaer, & O'Hara, 2002}; ^{Corwin & Schultz, 2012}). It can be applied to various studies in finance, including financial institutions (^{Ripamonti, 2020}). Thus, we deduce that such information can also relate to advertising and contribute to the marketing-finance interface.

The theoretical framework indicates that advertising actions would occur prior to the generation of value for the firm (^{Joshi & Hanssens, 2010}). Then, they are strategic activities whose information is treated by only the corporation’s managers (^{Akerlof, 1970}).

This research examines the relationship between the concepts of asymmetric information and advertising expenditure. For this purpose, we selected a sample of corporations from the developing stock market from 2007 to 2019, and the data were assessed using the panel cointegration technique (^{Larsson, Lyhagen, & Lothgren, 2001}). The results indicate that asymmetric information anticipates the generation of value due to advertising actions up to four quarters. It is consistent with the marketing-finance interface and contributes to understanding the enormous relevance of advertising actions as a substantive research topic (^{Edeling, Srinivasan, & Hanssens, 2021}).

Although research in the area of marketing-finance interface (^{Edeling, Srinivasan, & Hanssens, 2021}) has increased in recent decades-including the perspective on asymmetric information-this paper applies a specific measure to the developing stock market, filling part of a gap on the validity of international findings for this type of market, where a greater amount of information on advertising issues is expected.

This paper is structured as follows: the introduction section is followed by the theoretical framework, methodology, results, and concluding remarks.

2. Theoretical Framework

According to the advertising theory (^{Joshi & Hanssens, 2010}), the company’s value creation indicates the existence of direct and indirect effects of advertising on stock prices. This is consistent with the rational expectations theory (^{Muth, 1961}) and the efficient market hypothesis (^{Fama, 1970}; ¹⁹⁹¹). Therefore, planning advertising campaigns includes relevant strategic information (^{Akerlof, 1970}); it allows us to assume a long-term relationship (^{Johansen, 1988}; ¹⁹⁹¹) between asymmetric information and advertising, thus unfolding the market microstructure theory (^{Hasbrouck, 2007}).

Since high advertising expenditures are related to asymmetric information, our hypothesis is that this relationship is positive because asymmetric information impacts improvements on stock returns (^{Hasbrouck, 2007}).

3. Methodology

In general, panel data analysis studies predominate on the marketing-finance interface (^{Edeling, Srinivasan, & Hanssens, 2021}). However, some phenomena may show a lag among variables (^{Granger, 1981}), with the possibility of assesing the relationship between asymmetric information and advertising in the long term (^{Johansen, 1988}; ¹⁹⁹¹), especially when the effects can be seen over even longer periods (^{Joshi & Hanssens, 2010}). In addition, panel analysis of the data from non-stationary variables can lead to long-term spurious regressions (^{Granger, 1981}; ^{Engle & Granger, 1987}). Cointegration is the long-term relationship between non-stationary variables whose residual is stationary; it is adjusted in the short-term error correction models (^{Johansen, 1988}; ¹⁹⁹¹) applicable to panel data (^{Larsson, Lyhagen, & Lothgren, 2001}), as showed in Equation 1.

(1)

Among the measures used in the literature, selling expenses on the total assets were selected as the measure for advertising (^{Ding, Jia, Wu, & Yuan, 2017}), considering that since the internationalization of accounting standards, all advertising investments are recorded as selling expenses. For this reason, the sample period starts in 2007.

The asymmetric information is measured by S_2 (equations 2 to 5) or the Corwin-Schultz bid-ask spread estimator (^{Corwin & Schultz, 2012}), duly validated for the stock market of the sample companies (^{Ripamonti, 2016}). This measure requires only high (H), low (L), open, and close price data for its calculation, and its immediate availability determines the period of coverage of the sample until 2019.

(2)

(3)

(4)

(5)

The results are set up for growth opportunities set, size, leverage, and return using the traditional finance variables (^{Titman & Wessels, 1988}) of the M/B ratio (market value on book value of equity), size (ln of total assets), D/E (market value of debt on market value equity), and return (stock price changes). Consistent with the theoretical framework, this research assumes that there is a positive and significant relationship between asymmetric information and advertising.

The sample included data from 82 corporations listed on the Brazilian stock market during 57 quarters between 2007 and 2019. The most liquid stocks in this developing market in 2019 and the accounting of investments in advertising only for 2007 are adopted. Precautions in the analysis are necessary to avoid the survival biases and rejection of null cointegration that could be derived from the extension of the time series (^{Timmermann, 1995}). The survival bias could possibly be explained by the liquidity of the sample stocks. The data analyzed in this research were obtained from Comdinheiro.com financial platform.

The main descriptive statistics of the sample are presented in Table 1. It shows that the average S_2 was 0.0148 per day, while the average advertising was approximately 0.0485 for the assets. In general, relative consistency with other studies on the same market was observed (^{Ripamonti, 2016}), including control variables with an average return of 0.699 per quarter, M/B of 2.62, D/E of 0.68, and size of 10.25.

Table 2 shows the correlation between the variables assessed in this study, with no evidence of problems in the data. It indicates that advertising and S_2 have a negative correlation, contrary to the theoretical framework. This result may be due to a lag between the effects, which is confirmed by the reversal of direction in the coefficients of the short-term error correction mechanisms. Interestingly, the positive correlation between S_2 and D/E is consistent with the capital structure theories (^{Modigliani & Miller, 1963}; ^{Ripamonti, 2020}).

4. Results

The main result is the long-term relationship between advertising and asymmetric information, as shown in Table 4, with a lag of two quarters in all possible specifications of the applied technique. This result indicates that information on the development of advertising campaigns is captured by the measure of asymmetric information and confirms the relevance of this field of study (^{Edeling, Srinivasan, & Hanssens, 2021}).

One assumption of cointegration is the non-stationarity of variables, which is measured by the augmented Dickey-Fuller (ADF) test. Table 3 presents the stationary statistics of the examined variables through several tests, and it shows that all variables are stationary according to the statistics of the ADF-Fisher test (^{Dickey & Fuller, 1979}; ¹⁹⁸¹; ^{Larsson, Lyhagen, & Lothgren, 2001}). This would be the opposite of the cointegration assumption (^{Dickey & Fuller, 1979}), posing no problem when considering the existence of up to five cointegration posts, as shown in Table 4.

The optimal lag period for the relationship examined in this paper was two to eight quarters. For this study, we used the selection performed by the Schwarz information criteria, as it is more intuitive that the advertising investment effect that occurs in two quarters, although the long-term effect can occur in eight quarters.

The effective existence of cointegration is observed through trace and max tests, which indicate the number of possible cointegration ranks in the specified model. As shown in Table 4, the necessary convergence between the statistics was observed for at least five cointegration ranks. It confirmed the long-term relationship, even with the stationarity of the studied variables.

As predicted by the theoretical framework (^{Hasbrouck, 2007}; ^{Joshi & Hanssens, 2010}), the most significant variables for the relationship with asymmetric information were advertising and return in all types of trends for long and short runs (^{Larsson, Lyhagen, & Lothgren, 2001}), as shown in Table 4. Such a relationship has been observed in more stock returns (^{Du & Osmonbekov, 2020}), sales, (^{Guenther & Guenther, 2020}) and both (^{Bayer, Srinivasan, Riedl, & Skiera, 2020}).

There is an interesting result regarding capital structure. The results of all types of specifications indicate a significant relationship with the D/E variable, consistent with greater access to financing by companies with more advertising expenditures verified by ^{Ding et al. (2017)} . This evidence is also presented in ^{Kim et al. (2019)} .

The M/B variable, which measures the value of shares in relation to the book value of equity and demonstrates the set of growth opportunities for corporations, presents results consistent with other studies (^{Chemmanur & Yan, 2017}; ^{Garcia-Zambrano, Rodriguez-Castellanos, & García-Merino, 2018}). This is because investments in advertising preceding the initial public offering, as well as crises, generate lower returns in the long term. This is confirmed by the return variable, which measures the change in stock price, although there are short-term adjustments, as indicated by ^{Mian et al. (2018)} .

There are other comparisons for M/B and return va-riables. Results show a negative long-term relationship with sell_expenses/asset; thus, they are not consistent with the findings of other studies (^{Nguyen, Tran, & Baker, 2021}; ^{Kupfer & Schmidt, 2021}), which that indicated value creation and abnormal returns derived from advertising.

5. Concluding remarks

Based on the theories of value creation through advertising and market microstructure, this paper examined the possible long-term relationship between advertising and asymmetric information in a sample of corporations listed in developing stock markets over 57 quarters.

Planning advertising actions includes strategic information that is only owned by the managers involved in this activity. Thus, there is an evident difference in information between managers and investors that could be anticipated by an information asymmetry measure, specifically the one used in this study.

Since 2007, by virtue of the harmonization of international accounting standards, advertising investments have been recorded as advertising expenses. Therefore, it was assumed that the advertising variable was measured using the selling expenses to total assets ratio, and the sample’s analysis period started in the first quarter of the year.

Marketing-finance interface studies have mainly examined data through panel data analysis. However, this phenomena can occur at lags greater than those admitted by the aforementioned models. Therefore, we examined corporate data using cointegration panel techniques. The assumption was that asymmetric information and advertising would have a positive and meaningful relationship.

The main theoretical implication of this research is the confirmation of the long-term relationship between asymmetric information and advertising, adjusted by short-term mechanisms. Thus, advertising campaigns can be anticipated by up to two quarters using the asymmetric information measure employed in this study. Marketing-finance interface studies may consider that the long-term effect of advertising begins before the execution of campaigns. The effects on current sales, taken as short-term effects, can also influence the operation of corporations.

This opens up the opportunity to apply cointegration techniques and time series in general to marketing-finance interface studies, assuming that time is a relevant factor for the complete capture of corporate phenomena.

This study has many practical implications. For managers, it is possible to monitor the anticipated effects of the intended campaign and, eventually, make adjustments even before its execution. For investors, monitoring information asymmetry makes it possible to anticipate the strategic movement of corporations, closely determined by advertising.

The companies' compliance activities gain additional prominence since their operations can be monitored by investors to anticipate advertising actions and to rebalance positions in shares or derivatives.

Evidently, this research had restrictions related to the extension of the assessed sample and selected market data. It is also possible to replicate this study for developed stock markets that already registered advertising investments as expenses in accounting.