The impact of customer capital on company's market value: An empirical study from 100 U.S. stock market leaders

The difference between the market and book value of the company is largely due to its intellectual capital, usually grouped into three components–human, structural, and customer capital. Today, the issue of the impact of components of intellectual capital on the company's market value (MV) remains unresolved, which emphasizes the importance of this work.

In the effective value-based management of intellectual capital and its components for many companies, the main problem is the absence of opportunity for reporting information about intangible assets that would form a transparent analytical base for building current trends. In turn, the company's reporting, compiled in accordance with the requirements of international financial standards, provides an informational basis for the analysis of various types of intangible assets and allows determining the relationship between different indicators.

The importance of empirical studies on the impact of customer capital on the company's MV increases with the need to identify intangible reserves to build more effective management. Using econometric models built on the basis of available statistical data allows managers to establish the fact of presence and the level of dependence between various economic phenomena in the form of corresponding mathematical equations. Such an analytical assessment of data focuses on specialists on established trends in the past, making it possible to create a transparent resource management policy based on relevant forecasts.

The main motivation of this paper to prove the correlation between the customer capital and the company's MV. The result of studying the impact of customer capital on the MV should be an equation that allows you to understand how much the company's MV will change if the customer capital changes. Previous findings regarding the definition of the reflection features of intangible assets that characterize customer capital (IAcust) in the reporting allow us to confirm the vagueness and «fuzziness» of such data, which entails the need to highlight not one but several factors that contain information about customer relationships and are reflected in the reports. If there are a lot of factors, then we get correspondingly as many parameters (coefficients), with the help of which we can predict the company's MV.

Regression analysis requires a fairly large amount of calculations and verification of a number of hypotheses, but with the advent of appropriate software which simplifies the work of the analyst, free versions of such products can be used. In this study, we use such cross-platform software package as a Gretl: Gnu Regression, Econometrics and Time-series Library («GRETL»), designed for econometric modeling based on statistical data.

Knowing that customer capital is a component of intellectual capital, we have the opportunity to express how the problem of the impact of intellectual capital on various spheres of company activity is being revealed. The impact of intellectual capital on business organization includes such spheres as innovation, competitive advantage, research and development, business performance, value creation, organizational strategy, and competence and capability [Abdulaali, 2018].

There are researchers whose works are dedicated to the problem of establishing the impact of intangibles on the company's MV, in particular, factors such as intellectual capital [Chen et al, 2005; Tseng and Goo, 2005; Maditinos et al, 2011; Khangah et al, 2012; Berzkalne and Zelgalve, 2014; Bchini, 2015; Nimtrakoon, 2015; Nuryaman, 2015; Sardo and Serrasqueiro, 2017; Ahmed et al, 2019; Forte et al, 2019; Mačerinskienė and Survilaitė, 2019], intangible assets [Shahman, 2004; Volkov and Garanina, 2008; Behname et al, 2012; Ramirez and Hachiya, 2012; Shih, 2013; Castro, 2014; Jaara and Elkotayni, 2016; Glova and Mrazkova, 2018; Ocak and Findik, 2019; Vasconcelos et al, 2019], and goodwill [Chauvin and Hirschey, 1994; Shahman, 2004; Li et al., 2010; Cole, 2012]. Only one article related directly to the main focus of our study explores the relationship between customer capital and the company's MV [Taghieh et al, 2013]. These scholars establish the relationship between intangibles and the company's MV using mathematical models based on the reporting data from companies in different countries.

The problem of measuring the components of intellectual capital is polemic and difficult [Andreeva and Garanina, 2015]. The understanding that customer capital is the one component of intellectual capital in accounting is known as intangible assets leads us to the fact that researchers who devoted their works to the impact of intangible assets on the company's value could also study issues with the similar effect of customer capital (Table 1).

According to Chen et al. [2005] the advertising expenditure doesn’t have a significant impact on MV. In support of along this line of thought, Sardo and Serrasqueiro [2017] reject the hypothesis that the relational capital has a positive effect on firms’ MV. But the results in most articles (Table 1) show that the impact of customer (relational) capital on the company's MV is significant. Tseng and Goo [2005] propose eight useful value-creating paths, one of which is effectively managing the effect of relationship capital on enhancing corporate value. According to the results of Taghieh et al. [2013], relational capital has a significant and positive effect on financial performance and firm value. The authors conclude that customer capital, which is considered as a bridge or catalyst in intellectual capital activities, is the dominant and determining factor to change intellectual capital to the MV, accordingly, the company's business performance. Based on the survey data in Tunisia, Bchini [2015] finds that the link between relational capital and value creation is linear and positive in manufacturing companies. The analysis of Yilmaz and Acar [2018] reveals that relational capital efficiency (RCVA) has a significant positive impact on the market to book (M/B) ratio, which is used for evaluating the MV of companies. Supposedly, the authors’ conclusions differ because of the selected data were taken from different countries’ reporting.

In this study, research and development (R&D) costs (expenses, expenditures) are also considered as a factor that includes information about customer capital and the level of investment in customer relationships. There are articles devoted to the impact of R&D costs on the company's MV [Chen et al, 2005; Lantz and Sahut, 2005; Duqi and Torluccio, 2011; Ramirez and Hachiya, 2012; Glova and Mrazkova, 2018; Pazarzi and Sorros, 2018].

From the result of the literature review, we find only one work in which the problem of the impact of customer (relational) capital on the company's MV is investigated [Taghieh et al, 2013]. Therefore, such empirical research becomes necessary for the development of a direction for studying the importance of managing this type of capital.

We hypothesize that customer capital has a direct impact on the company's MV. Following this we test the relation between customer capital and the company's MV using three variants of the regression model (Table 2). The models are the same in the part of the dependent variable, which is represented by the MV, and differ in the number of independent variables.

IAcust are calculated as the sum of such intangibles as brands (brand assets), customer base, customer contracts, customer lists, customer programs, customer relationships, customer-related assets, costs incurred to obtain contracts with customers, loyalty card holders, order backlog, purchasing and payer contracts, trademarks, and trade names. Other intangible assets are calculated as the total intangible assets, minus the total IAcust. GW, research and development costs and TA are taken from annual report (Form 10-K) of analyzed companies. The level of companies’ research and development intensity is determined according to results of work of Galindo-Rueda and Verger [2016], who proposed the classification of R&D intensity by industry. FinLev is calculated by dividing the total liabilities by the shareholder equity of the analyzed company.

The first variant of the regression model (Model 1)—simple linear regression—consists of the dependent variable (MV) and the independent variable (IAcust). This model is built by using the cross-platform software package «GRETL», and follows the regression Equation (1.1) stated as follows: (1.1) y^=135010+1.46x1, \hat y = 135010 + 1.46{{\rm{x}}_1}, where

y—MV (market value);

The failure of Model 1 is caused by a low coefficient of determination (R²) of value 0.0045. The closer coefficient of determination (R²) is to 1, the more real the results of using the model will be.

This ratio shows what part of the dependent variable is explained by regressors. The determination coefficient varies in the range of [0, 1] and should be >0.2 [Mačerinskienė and Survilaitė, 2019].

It is inadvisable to use the regression equation of Model 1 for predicting the company's MV due to the lack of links between it and its customer capital. According to the results of the analysis, using a simple linear regression with one factor based on the reporting data of 100 U.S. stock market leaders for 2018, we came to the conclusion that the intangible assets characterizing the customer capital do not affect the company's MV. The problem of the lack of links between these measures can lay in the issue that IAcust does not demonstrate the real size of customer capital.

According to Yilmaz and Acar [2018], the models with multi-factor components present better results in understanding MV. Therefore, the inappropriateness of Model 1 leads to the need to expand the number of factors characterizing customer capital, which hypothetically can make the model more correct for practical users. Though we know the number of companies that don’t have IAcust in their reports, we have to take into account that the absence of such reporting data in 35 out of 100 companies does not prove their actual absence. Thus, due to the failure of the single-factor model, we will use regression analysis with other factors that can characterize the customer capital of the company.

In the second variant of the regression model (Model 2) we use the company's MV as a dependent variable, and as independent variables we use IAcust, other intangible assets, GW, and R&D costs.

This model, built using the cross-platform software package «GRETL», follows the regression Equation (1.2): (1.2) y^=65729.5+0.71x1+0.42x2+0.11x3+26.79x4 \hat y = 65729.5 + 0.71{{\rm{x}}_1} + 0.42{{\rm{x}}_2} + 0.11{{\rm{x}}_3} + 26.79{{\rm{x}}_4} where

y—MV (market value);

According to the t-test results, the constant (const) and the R&D costs (RD, x₄) are recognized as the most significant in the presented regression equation.

Table 3 shows the statistical characteristics of the sample.

Table 3 shows summary statistics for all variables of the analyzed regression model (Model 2), which generally inform about the values from the sample (e.g., minimal, average (mean), and maximum value for each variable). It should be noted that for all independent variables the minimum value is 0, which means in the sample of companies there are companies that do not have reporting data for the selected factors that characterize the customer capital.

In support of the practicability of Model 2, we note that the coefficient of determination (R²) of that model is 0.4, which is higher than in Model 1 (R² = 0.0045). This makes it possible to use this equation to predict the company's MV with a probability of 40%.

The presented regression (Model 2) of 40% explains the changes in the company's MV (MV, y). The remaining 60% of the changes in the MV are due to other factors that are not included in the presented equation as independent variables.

According to the results of F-test (Fisher test), the proposed regression model (Model 2) is recognized as adequate to sample data, because the comparing (using the cross-platform software package «GRETL») of the observed value of F [4, 95] with the critical value of F [4, 95] shows that F_observed is higher than F_critical (F_critical (2.47)<F_observed (15.76) with a probability of error of 0.05, which confirms the basic requirement of this test.

One of the requirements of multiple regression is the lack of highly correlated independent variables. First, we demonstrate the correlation coefficients for the independent variables of Model 2 (Table 4).

The correlation coefficient between GW and IAcust is 0.65, which is the most significant correlation between independent variables in Model 2.

Next, we will construct a correlation matrix to establish the level of the collinearity of the independent variables of Model 2 (Figure 1).

Figure 1

The data from Table 4 and Figure 1 suggest the absence of multicollinearity of the variables of Model 2, due to the lack of correlation coefficients with a value >0.7. There is one negative correlation between GW/IAother, which means that increasing one variable (GW) will decrease another (other intangible assets).

For another variant of testing for multicollinearity, we estimate the regression models, where x₁, x₂, x₃, and x₄ will be dependent variables, and all other factors will be independent variables. Thus we find the coefficients of determination of variables (Table 5).

According to the data of Table 5 the multicollinearity for the selected variables is rejected because all values are <0.7.

The proposed regression equation of Model 2 can be used by managers to predict the company's MV with a probability of 40%. Verification by various tests shows that model is adequate to sample data (F-test) and the multicollinearity for the independent variables is rejected (correlation matrix, coefficients of determination).

Understanding that about 60% of changes in MV are due to other factors that are not included in the presented equation (Model 2) as independent variables, we expand the number or the independent variables to increase the probability of predicting MV.

In the third variant of the regression model (Model 3) we use the company's MV as a dependent variable while IAcust, other intangible assets (IAother), GW, R&D costs (RD), the company's size by its TA, the Int, and FinLev as independent variables.

Now add to Model 2 such independent variables as the company's size, the Int, and FinLev, which are used to control the industry and other specific features of the company. We assume that the added factors have a significant effect on the company's MV.

In particular, the size of the company is a very important indicator that directly affects the MV of the company. The reporting makes it possible to see the size of the company as TA, but in calculations we use the natural logarithm of TA at the end of period to reduce the dispersion of data and simplify their processing.

Using the findings of the work of Galindo-Rueda and Verger [2016, p. 14], who classified companies by its average weighted levels of R&D costsand the specifics of the company's activities into levels, we divide 100 U.S. stock market leaders in accordance with this taxonomy (high, middle, or low R&D intensity). These authors relate the ratio of R&D intensity to value added within an industry.

FinLev is a ratio calculated by dividing the company's total liabilities with the company's shareholder equity. The information about this ratio is necessary for investors to understand the level of possible risks.

This model, built using the cross-platform software package «GRETL», follows regression Equation (1.3): (1.3) y^=−794438+0.24x1−0.9x2−1.45x3+19.55x4+80992.2x5+12414,4x6−21.51x7 \hat y = - 794438 + 0.24{x_1} - 0.9{x_2} - 1.45{x_3} + 19.55{x_4} + 80992.2{x_5} + 12414,4{x_6} - 21.51{x_7} where

y—MV (market value);

According to the t-test results, the constant (const), R&D costs (RD, x₄) and the TA are recognized as the most significant in the presented regression equation.

Table 6 shows the statistical characteristics of the sample.

The difference between Model 2 and Model 3 is the addition of three control variables (TA (log), Int, FinLev) in Model 3. The Table 6 shows summary statistics, which generally inform about the values from the sample (e.g., minimal, average (mean) and maximum value for each variable).

According to the calculations in the cross-platform software package «GRETL», the coefficient of determination (R²) of Model 3 is 0.52, which is higher than in Model 1 (R² = 0.0045) and higher than in Model 2 (R² = 0.4). This makes it possible to use this equation to predict the company's MV with a probability of 52%.

According to the results of F-test (Fisher test), the proposed regression model (Model 3) is recognized as adequate to sample data, because the comparing (using the cross-platform software package «GRETL») the observed value of F [7, 92] with the critical value of F [7, 92] shows that F_observed is higher than F_critical (F_critical (2.11)<F_observed (14.09) with a probability of error of 0.05, which meets the basic requirement of this test.

Next we check the correlation coefficients of the independent variables of Model 3 (Table 7).

The correlation coefficient between GW and IAcust is 0.65, which is the most significant correlation between independent variables in Model 3 (similar to Model 2). The correlation matrix, which establishes the level of the collinearity of the independent variables of Model 3, is shown in Figure 2.

Figure 2

Similar to Model 2, the data from Table 7 and Figure 2 suggest the absence of multicollinearity of the variables of Model 3, which is due to the lack of correlation coefficients with a value >0.7. There are some negative correlation values which mean that increasing of one variable will decrease another (e.g., IAcust/Int, IAother/RD, FinLev/TA).

For another variant of testing for multicollinearity, we estimate the regression models, where x₁, x₂, x₃, x₄, x₅, x₆, and x₇ will be dependent variables, and all other factors will be independent variables. Thus we find the coefficients of determination of variables (Table 8).

According to the data of Table 8 the multicollinearity for the selected variables is rejected because all values are <0.7.

Model 3 can be used by managers to predict the company's MV with a probability of 52%, which is higher than in Model 1 and in Model 2. Verification by various tests shows that model is adequate to sample data (F-test) and the multicollinearity for the independent variables is rejected (correlation matrix, coefficients of determination).

The proposed variants of regression models prove that the customer capital as intangible assets characterized customer capital (the sum of «Customer-related intangible assets», and partially «Marketing-related intangible assets») has a direct impact on the company's MV, but the values of this impact are insignificant.

It should be noted that with an increase in the number of factors in the regression equation, the value of the impact of the customer capital on the company's MV decreases. The regression coefficient of the customer capital in Model 1 equals 1.46 (1 factor), in Model 2—0.71 (4 factors), in Model 3—0.24 (7 factors). This coefficient describes the relationships between customer capital and the company's MV. The positive sign indicates that as customer capital increases, the company's MV also increases.

All variants of regression models include information about intangible assets that characterized customer capital, but not all can be recommended for practical use. We don’t recommend Model 1 as it is rejected for practical use because of its low value of R² (0.0045). This statistical measure (R²) determines the proportion of variance in the dependent variable (company's MV) that can be explained by the independent variable (customer capital). In turn, the coefficient of determination (R²) in Model 2 is 0.4, which allows managers to use this regression equation to predict the company's MV with a probability of 40%. But the highest probability is ensured by the use of Model 3 in which the coefficient of determination is 0.52.

Various tests conducted by us show that Model 2 and Model 3 are adequate to sample data (F-test) and the multicollinearity (for the independent variables) is rejected (correlation matrix, coefficients of determination). Therefore, Model 2 and Model 3 are recommended for practical use.

The aim of this paper was to establish the relationship between the customer capital and the company's MV. The hypothesis of this study is the fact that the customer capital has a positive impact on the company's MV. Similar issues were developed in previous research by other authors, in particular, the works of Chen et al. [2005], Tseng and Goo [2005], Taghieh et al. [2013], Bchini [2015], Sardo and Serrasqueiro [2017], and Yilmaz and Acar [2018]. But it is noteworthy that the authors’ conclusions are not the same in the context of the impact of customer (or its components) on the value of the company.

The unique aspect of the study was the establishment of the impact of customer capital on the company's MV from two perspectives. The first approach provided that customer capital represented only IAcust. According to the requirements of IFRS 3 «Business Combinations» we combined «Customer-related intangible assets» (customer lists, customer relations, etc.) and partially «Marketing-related intangible assets» (only trademarks, brands) that constituted the object of this approach (IAcust). In the second approach, the information about customer capital, in addition to the aforementioned intangible assets, could be dispersed to other intangible assets, GW and R&D costs.

The study demonstrates that the first approach (Model 1) is rejected for practical use, which is due to the low value of the coefficient of determination. In turn, the coefficient of determination in the second approach is higher and allows managers to use the regression equation to predict the company's MV with a probability of 40% (Model 2) or 52% (Model 3).

Model 3 (as an extended version of Model 2) demonstrated a higher probability in predicting the company's MV than Model 2, due to the addition of control variables such as the company's size, the Int, and FinLev, which are used to control the industry and other specific features of the company.

The presented regression analysis based on the reporting data of 100 U.S. stock market leaders showed that intangible assets characterizing customer capital have a direct impact on the company's MV, but the values of this impact are insignificant. Therefore, we recommended the impact of customer capital on the company's MV analyzed using different indicators, which may include information about customer capital. Particular attention should be paid to R&D costs, which have a significant impact on the company's MV according to the findings of the regression analysis.

This study has some limitations. One of the limitations of this study is the relevance of the research results can’t be applied to all companies, but only to 100 U.S. stock market leaders. These companies are exemplars, which leads them to maintain the most transparent information policy that provides open access to reports that are provided as quickly as possible, which affects their position in the stock markets. Further development of research in this direction may cover other samples of companies, which will reveal general trends and differences in results. The second limitation of this study is taking into account information only for 1 year (2018). Therefore, one of the possible directions for future research could be a comparison of results of identical regression models in dynamics (for example, for the period of 5 years), which allows researchers to identify trend changes in development.

The needs of value-based management require the establishment of relationships between various factors and the company's MV in the form of mathematical models. A priori the intangibles are very important in the process of creating value for companies. The developed models can be used as a guide for managers in the processes of effective management of intellectual capital and its components, which enables seeing links between the company's MV and intangible value drivers.