Intangible-intensive strategy in crisis

The last type of intangibles is relational capital. It stands for external network and relations with partners and stakeholders, including mass media and government. In this study relational capital is represented by amount of citations, employment of foreign capital and participation in business associations. The first structural indicator is membership in business associations. As a participant of business association companies can establish communication with competitors, business partners and these associations provide platforms for exchange of experience. It is can be useful for improving an actual position in the market [Acuna et al,2012]. Indicators are binary variables, based on companies' Annual Report. The next variable is amount of citations in search engines. This indicator shows the level of recognition of the company among its stakeholders and primarily customers. [Shakina and Barajas, 2012] have shown the effect of this indicator on the creation of a company's value. The indicator represents citation in search engines based on Google PageRank (0-10). One more variable that accounts to structural capital is foreign capital employment. The variable was used in the study of Shakina and Barajas [Shakina and Barajas, 2012] as intellectual capital component for assessing influence on value creation of companies. Foreign capital provides opportunity for attracting additional funding to the company. It is a binary value, based on information from Annual Report about shareholders. Presence of access to foreign investments will be indicated by the presence of a company of foreign citizens as investors. The indicator reflects presence of foreign investors in the company.

Throughout this study dependable variables for the investigation are the economic value added (EVA) and the market value added (MVA), which reflect performance of the company. These indicators were in detail reviewed and compared in the Theoretical background section of the study. The control variables for the regression analysis will be book value of the company, which reflects the indicator of size of the company and company age for reflection of previously accumulated intellectual resources. Summary of detailed descriptive statistics for all variables for the samples of European and Russian companies is in the Table 3 and Table 4. All numeric variables of the samples are not normally distributed according to the descriptive statistics (Kurtosis indicator > 3; Skewness indicator is not in the area (-2; 2) for both samples of European and Russian companies.

Table 3. Descriptive statistics for the sample of European companies

Table 4. Descriptive statistics for the sample of Russian companies

Correlation analysis for chosen variables of intangibles was conducted for the samples of European and Russian companies. The analysis discovered weak correlation between most of the variables. In the sample of European companies positive moderate correlation exists between variables of ERP system implementation and corporate university (42 %); corporate strategy and ERP system implementation (40 %); citations level and intangible assets (28%) (Table 5). In the sample of Russian companies, positive moderate correlation can be observed between variables of corporate strategy and ERP system implementation (41 %); intangible assets and corporate university (24 %). Overall, these indicators are not critical for the further investigation.

Table 5. Correlation analysis of intangibles variables for the sample of European companies

Table 6. Correlation analysis of intangibles variables for the sample of Russian companies

In order to test the above hypotheses and reach objectives and aim of the thesis it was conducted investigation of public companies from panel data during three periods: before the economic crisis (2004-2007), during the crisis (2008-2009) and after the crisis (2010-2013). Same analysis is conducted separately for companies that operate in European market and for those that work in Russia and comparative analysis is conducted.

First step of the research is based on dividing companies on intangible-intensive non-intensive profiles with defined variables of intangibles in three periods: before, during and after crisis. The methodology of defining intangible-intensive profile of company was mentioned in the paper of [Barajas, Shakina & Fernбndez-Jardуn, 2017]. The dummy variables for intangible-intensive profiles of companies in human, relational and structural capital were created by MS Excel program with functions “IF”, “AND” and “OR”. The theoretical framework of [Molodchik, Shakina, & Barajas, 2014] was used for creating variables. The main criteria for defining intensive profiles was in that the company has higher indicators of intangibles than the average and have special intangibles in the cumulative level, which doesn't have other companies. The company with intensive profile in human capital has higher level of qualification of board of directors (more than median) and invests more on each employee (more than median), has higher earnings from employee (more than median) or provides corporate university. The company with intensive profile in relational capital has more citations on search engines (more than median) and have foreign capital employed or have membership in business associations. In addition, the company with intensive structural capital has intangible assets more than median and has also ERP systems, corporate strategy and patents more than median (Table 7).

Table 7. Criteria for intangible-intensive strategy

To answer the first question of the study: whether being intangible-intensive helped European and Russian companies overcome global crisis of 2008-2009 with less decrease in performance (measured in EVA and MVA) and recover faster than non-intangible intensive ones; for testing the hypothesis related to the question medians of the EVA and MVA indicators, which reflects performance of the company, are compared between intangible-intensive and not intangible-intensive companies before, during and after crisis. Nonparametric Mood's median test is used for comparison of medians of the samples with usage of Stata program package. The study uses median test, because according to the descriptive statistics, numeric variables are not normally distributed and variables have outliers. Such actions as generating variables of crisis periods, conducting median test in Stata with chi-square test and exact function with a Fisher test for the samples with the intangible-intensive and not intangible-intensive companies in the periods related to crisis are done. As for the better interpreting results the medians of both samples in the crisis periods are also presented.

As in the study of [Shakina and Barajas, 2012] this research selected an approach with the assessment of intangibles based on the value-based and hedonic pricing approach. The value-based approach allows determining the numerical characteristics of proxy indicators of intangibles, exploring more about the investment preferences of the companies and interpret findings about a particular component of intellectual capital. Hedonic pricing approach evaluates the added economic value of a company as a result of the utilization of the company's intangible resources and makes it possible to estimate the total effect of all intangible resources. According to this, in this study the input parameters of the intangibles are evaluated as explanatory variables in the regression model, and the outcome is the economic and market value added.

The study is concentrated on the intangible-intensive companies, and therefore regression models will be built for the sample with intangible-intensive companies for answering the second research question: how specific elements of human, relational and structural capital influence on the performance of European and Russian intangible-intensive companies before, during and after crisis, and its sub-question about the presence of the industry effect for the significant intangible-resources. As the sample is presented in the panel data, the appropriate for the study model will be model with fixed effects. Equation for the model with fixed effects is following [Brьderl & Ludwig, 2015]:

Yjt = вxjt + бj + ujt (3)

where: Yjt- outcome of observed entity j at time t, dependent variable,

бj (j=1….k) - individual unkown intercepts (fixed for j) ,

xjt - independent variable (j=1….k at time t),

в - the coefficient for the independent variable,

ujt - the error term вj.

The usage of model with fixed effects allows controlling individual time-invariant effects of companies. This model is suitable for the research also, because variables changing within time periods are analyzed; the model gives the net effect of the repressors on the dependable variable Dependent variables will be presented by EVA and MVA indicators, which reflects performance and were described in previous sections. These performance indicators will be analyzed in separate models. The main steps for answering research question contain generating each chosen intangible variable for intangible-intensive companies before, during and after crisis; then building regression for EVA and MVA with independent variables of intangibles for defined sample of intangible-intensive companies and periods. Within the research the regression models will be built for European and Russian companies separately and main trends will be compared further. The results of these regression models answers the first sub-question and test hypothesis related to this question.

For the second sub-question: whether significant intangibles influence on the performance of intangible-intensive companies with predominance in one industry, the results of the first model will be used. The model will be applied for the significant intangibles in all observed period with effect of industries.

Models of the study for answering research questions can be expressed with following functions:

СP = f (Int, IS, Pc, CV) (4)

СP= f (Int1, IS, Pc, Ind, CV) (5)

where: CP - Company performance indicator (EVA, MVA)

Int- intangibles of human, relational and structural for all companies

Int1 - significant in all periods intangibles (from initial model)

IS -applied intangible-intensive strategy for companies

Pc- periods (before, during, after crisis) for the intangible-intensive strategy

Ind- effect of industry

CV- control variables.

The regression models were built in the Stata statistical software package, which allows analyzing panel data; and such commands as generating variables for investigating additional conditions for the variables and building regression model with fixed effect were used.

4. Description of the results

Comparative analysis of medians for EVA and MVA.

The first question was whether intangible-intensive companies showed higher performance in terms of MVA and EVA during, before and after crisis. To answer the question, we conducted comparative analysis of medians for intangible-intensive and non-intangible-intensive companies. Median was chosen because this method of measuring average is more objective in case when the distribution is not normal. The research was conducted separately for Europe and for Russia. First, results for Europe were described.

The first hypothesis was that EVA and MVA for non-intangible intensive and intangible-intensive companies differ before, during and after crisis. To test this hypothesis median test was conducted.

To test whether distribution for EVA and MVA in the sample was normal the study observed kurtosis indicator from the detailed descriptive statistics. For normal distribution kurtosis varies close to 0, distribution can be considered normal if the kurtosis indicator is from 0 to 3. In the case for EVA kurtosis equals 122, for MVA it is 133, therefore in none of the cases distribution is normal. That is why using simple average would be incorrect and median was chosen for estimation.

To test statistical significance for comparing medians Fisher's exact and Pearson chi2 were used. For analysis the study focused on Fisher's exact. Medians were compared for each type of intangibles (human, relational, structural) for each of three periods (before, during, after crisis). For EVA by human capital during crisis Fisher's exact equals 0.075, one-sided Fisher's exact is 0.040.

The same after crisis is 0.027 for Fisher's exact and 0.014 for one sided. For all the other periods and intangibles Fisher's test showed 0, for both EVA and MVA.

This means that there is significant difference for median EVA and MVA for each group of intangible-intensive companies versus non-intangible intensive ones. EVA and MVA are somehow influenced by the intangibles.

The tables below (Tables 8 -11) shows the results for the median tests for EVA and MVA for the samples of European companies.

The distribution shows that all the groups are appropriate for comparation in size. The test also explains the skewness of the median and average.

Table 8. Median test for EVA, Europe*

Table 9. Median test for MVA, Europe

The medians for EVA and MVA are presented in the tables below. “Median if 1” in the table means the median EVA for intangible-intensive companies, “Median if 0” stands for the others. Column “Difference” was added to show the difference between medians “1” and “2”. Fisher's test results are shown in the column “Difference”: results that are significant on 99% confidence level are marked ***, 95% are marked **, 90% are marked *, others are not marked.

Table 10. Medians and significance for EVA, Europe

Table 11. Medians and significance for MVA, Europe

The results show that median EVA and MVA for intangible intensive and non-intangible intensive companies were different. The difference was significant on 99% confidence level for MVA and at least at 90% confidence level for EVA. Therefore, for Europe H1 can be accepted.

Analysis of medians showed that during all three periods average EVA was negative for all the groups of companies. Human capital intensive companies showed less decrease in EVA before, during and after crisis. Relational and structural capital intensive companies, on the contrary, showed much lower average EVA for all the periods, especially for the crisis period of 2008-2009. Average MVA, at the same time, was much higher for intangible-intensive groups for all periods. The largest difference in MVA was observed before crisis for relational and structural capital intensive companies. During crisis there was a sharp decline in MVA for the groups, but MVA remained positive for intangible intensive companies, while for relational and structural capital non-intensive companies MVA dropped to the negative positions in this period. After crisis MVA continued to grow much faster for companies with relational and structural capital. Though for non-intangible intensive companies EVA was lower, MVA for them was much higher.

Based on the above analysis, the study finds out that MVA and EVA show different tendencies for the same groups of intangible intensive and non-intensive companies. These results fit to the statement, provided in previous research, that in case of negative EVA correlation between EVA and MVA is weak. The researches used to explain this phenomena by the high loyalty of shareholders towards large reputable firms (even in case their EVA for the period is negative) and by the role of large capital of the firms. Our research now is going to test whether intangibles are influential for growth of market value of the companies during their economic value recession.

The next step was to conduct the same analysis for Russian companies. Kurtosis for Russian companies equals 399 for EVA and 70 for MVA, so the distribution is not close to normal. Median was chosen as the indicator for average. The results for the median test for EVA and MVA are presented in the tables below (Tables 12-15). The same as for European companies, all the groups are appropriate for comparation. Some skewness of the average and median was also observed.

Table 12. Median test for EVA, Russia

Table 13. Median test for MVA, Russia

The medians for EVA and MVA are presented in the following tables. The same methodology as for European companies was used.

Table 14. Medians and significance for EVA, Russia*

Table 15. Medians and significance for MVA, Russia

For Russia not all the groups showed significant difference in medians according to the Fisher's exact. That is why H1 cannot be accepted for all periods for Russia.

Russian intangible-intensive and non-intangible intensive companies did not show such a great difference in average performance as European. Average EVA for all the periods for non-intangible intensive companies was negative. However, for intangible-intensive companies positive average EVA was observed only for human capital intensive companies before crisis. In general, EVA for intangible-intensive crisis was lower than for the rest, except for human capital before 2010, but the difference was not big. As for MVA, intangible-intensive companies performed better before crisis, but the value fell significantly in crisis and did not fully recover till the end of the observed period. During crisis human capital intensive companies on average overperformed non-human intensive ones, but relational and structural capital intensive companies already had lower MVA then competitors. After the crisis period the companies did not reach previous results and even showed the results much lower than in 2008-2009. That refers both to intangible intensive and non-intensive companies. This dropdown indicates that in reality crisis period did not end in 2009 for Russia, economic recession was continuous and global crisis of 2008-2009 for Russia lasted much longer and was followed by local crisis of 2014. But that already stands out of the focus of the research.

Observation of European companies figured out a “strange” trend: in some periods intangible intensive companies showed much lower EVA than non-intangible intensive, but MVA remained much higher. It was decided to observe this trend deeper to figure out whether such phenomena was related with the companies' strategy.

Usually positive EVA shows strong correlation with MVA, and then the common explanation is used (EVA is the main factor, defining MVA, as market price of the shares is dependent on the amount of money on which the company's profit exceeds its investments). However, in case of negative EVA, very weak correlation between the two variables was observed [Kramer & Peters, 2001]. In some cases negative EVA did not lead to sharp decline in MVA. MVA did not change or even continued to grow. Such phenomena was explained by high loyalty to large reputable companies and large accumulated capital. However, investments into intangibles might give alternative explanation. The suggestion is that companies, developing intangibles, cannot quickly reinvest the money in crisis and cover current losses, but the investors consider intellectual capital as part of stable capital and involve it to the value of the shares as they expect future benefits of these investments.

To answer the question, first of all, companies with falling EVA and growing MVA were figured out. For that growth in EVA and MVA was calculated for all years, except for 2004. The new sample included 3110 rows (each row represents a company's annual performance). Out of them 1895 refer to intangible intensive companies (a company was called intangible intensive if it fitted criteria for at least one group - human, relational or structural capital intensive. Some companies were included to several groups). Among them were 439 human capital intensive companies, 1249 companies with relational capital and 1649 with structural. As the pie chart below illustrates, the majority of such companies turned out to be intangible-intensive. Moreover, the largest part was formed by companies with structural capital intensive strategy.

Figure 5. Companies with growing MVA and falling EVA

Figure 6. Intangible-intensive companies with growing MVA and falling EVA by type of capital

The next step was to test whether there is correlation between intangible-intensive strategy of the companies and their positive MVA despite negative EVA. To test it Pearson function was used. However, the results showed only weak correlation: for intangible-intensive companies in general it was 6%, for structural capital it was 11%. Therefore, for Europe there were more intangible-intensive companies with growing MVA despite decline in EVA than non-intangible intensive ones; the correlation was positive, but weak.

Impact of intangibles on the performance of Russian and European intangible-intensive companies in the period of changing economic cycles.

To answer the second research question and its first sub-question it was necessary to build fixed effect models for the performance of the company and independent variables as indicators of human, relational and structural capital with effect of crisis related periods and having intangible-intensive profile as it was mentioned in the previous section of the thesis. Crisis related periods were before crisis (2004-2007), during crisis (2008-2009) and after crisis (2010-2013). The dependable variables were EVA and MVA. The independable variables are human capital indicators as qualification of board of directors, corporate university presence, earnings and costs per employee; relational capital indicators as foreign capital employment, citations in search engines and membership in business associations; structural capital indicators as intangible assets, number of patents, corporate strategy implementation and ERP system implementation. The control variables for the model are book value of the company's assets for reflecting its size and company age for reflecting accumulated knowledge by the organization. The models within the study were built for the sample of Russian and European companies independently. Table 16 presents results of the regression model with EVA outcome for Russian and European companies. The number in brackets refers to standard error. Results that are significant at 99% confidence level are marked ***, at 95% confidence level **, 90% *. Results without marks are not significant at 90% confidence level a...