FACTORS THAT INFLUENCE THE SUCCESS OF SMALL AND MEDIUM ENTERPRISES IN ICT: A CASE STUDY FROM THE CZECH REPUBLIC

. Small and medium enterprises (SMEs) occupy a large niche in the information and technology sector (ICT) and play an important role in the functioning on any state’s economy. This paper focuses on a specific local market and uses the Czech Republic as a case study in order to establish the success factors crucial for achieving economic success. It aims at determining those factors with the help of econometric success rate models based on the own data collected via the means of questionnaire survey among ICT enterprise. Our results show that the earnings­employee ratio, average revenues and the investment in own R&D play the most important role in the success of Czech SMEs in question. Both, financial and non­financial indicators perform significantly in the predictions of success.


Introduction
Small and medium enterprises (SME) represent a relevant field of research because of their significant role in every economy (see e.g.Lukács 2005;Kočenda 2008;or Tech, Latha 2012).Based on a study conducted by the Ministry of Industry and Trade of the Czech Republic (MPO), on December 31 st 2011 there was 1,066,787 SMEs that formed 99.84 percent of all registered and active enterprises in the Czech Republic.Their significance underscores the fact that more than 60 percent (1,856 thousand employees) of Czechs are employees of SME, their performance ratio in 2011 cre ated 49.5 percent of the ICT sector (CZK 4,064,795 million), and performance value added represented 54.43 percent with a value of CZK 1,342,297 million (MPO 2011).
The Czech Statistical Office (CSO 2012) declares the ICT sector on the basis of CZNACE and OKEČ classification; for the purpose of this work we use CZNACE.The ICT sec tor comprises all economic entities (legal persons, natural persons) whose principal activities include the following compare the development of employment in ICT with the whole Czech labor market (CSO 2012).It is noticeable that during the observed years, except for the deepest pe riod of "financial crisis" (2008)(2009), there was moderate growth in job position creation.Moreover, the share of ICT job places on the whole market rose as well.The slopes in Figure 1 denote number of people employed in ICT sector (thousands of people).Lin (1998) denotes the number of employees in ICT sector as percentage of employees in all the sectors (CSO 2012).
It is obvious that the emerging importance of ICT could not be described by just the growth of employment in the sector alone.The World Bank has developed a separate unit to coordinate investments in the ICT sector.The rationale behind this is that technological progress is a considerable driving force behind economic growth.ICT infrastructure in particular has attracted much investment, and generated significant fiscal revenues and employment opportunities in developing countries (see e.g.Zálešák, Kocourková 2009).The number of mobile phone subscriptions in developing countries has increased from 200 million in 2000 to 3.7 bil lion in 2010, and the number of Internet users has grown more than tenfold (CSO 2012).
More specific details for local conditions are given in research encompassing 50 ICT companies in 2009.A survey showed that 80 percent of companies consider the reces sion to have run its course by the end of 2009 (Zálešák, Kocourková 2009).To extend this idea, Voříšek and Novotný (2010) state that the Czech Republic's ICT sec tor is still developing and we could consider it as the most progressive among with the financial and energy sectors in the past decade.To support this statement, Figure 2 below shows the production value of the ICT sector and its com parison with the entire business enterprise sector.As we can see, the production grew from CZK 73 billion over the past 18 years to CZK 522 billion, and even the proportion in production of every business enterprise nearly tripled, from 2.3 percent to 6.1 percent.
Overall, SME in the Czech ICT sector plays one key role.They definitively form the organizational structure related to the amount of companies in this sector.Furthermore, more than 60 percent of legal personalities were employed during the past 8 years (around 1.6 percent of total em ployment) in this sector.The financial crisis had a strong impact on ICT SME, with the number of enterprises trend ing downward as of 2009, while the rest of them started to invest more in R&D to implement new technologies and products.However, ICT shows signs of competitiveness and it is still possible for new enterprises to enter the market.This is because of the fact that over the period of 2005 to 2011, the number of SME grew while their revenues rose, but the profit margins were trimmed back.In the case of the Czech Republic, ICT sector will perform as a prosperous and absolute core of business.The number of workers in information technology is slowly approaching 200 thou sand, which is almost equivalent to the automotive industry (Voříšek, Novotný 2010).Moreover, it is believed that most new investment will come to ICT.Therefore, supporting development of the bio and nanotechnology fields is crucial.
The main objective of this paper is to discover and describe the most significant factors, and indicators, or rather even methods, that influence the success of SME in the Information and Communication technologies sector (ICT) in the Czech Republic.This sector was selected be cause it is currently undergoing a revolution in its develop ment with public demand and business opportunities for adaptation of new products and ideas enabling companies to achieve rapid market penetration.However, for their stable development there is a further added value needed.This work focuses on the factors that companies consider to be important.

Definitions of SMEs success
Motivation for describing the success or growth of SME comes from various sources all around the world.The main reason represents the research dedicated to improvements and advice to the government.As was mentioned above, Business: Theory and Practice, 2015, 16(3): 304-315 SME creates the majority of the companies established in the given state, thus the implementation of governmental policies should not be harmful to the environment for wealthy enterprise growth.Kaibori (2001) emphasizes the importance of such positive national conditions, especially among transition countries.Meanwhile, the entrepreneurs have come to establish enterprises on their own and with their own capital in order to secure adequate income and stable working conditions for em ployees.Government usually faces a strong decline in working conditions and economy.It is crucial for them to set appropriate rules that would create smooth and stable growth conditions.
In this kind of environment, it is necessary for own ers to establish their companies with individual resources like quality of service or goods, capital, sales, management, knowhow, technology, information, business connections, and then accumulate them and expand into business oppor tunities.This usually creates a very important environment for national competitiveness and economic growth.
One could argue that these indicators, which need to be researched for policy setting reasons, are not common and that the diversity in culture, social conditions, historical background or stage of economic development among the individual states in the world (see e.g.Kadocsa, Fransovics 2011).Kaibori (2001) oppose with the suggestion that "there are many things in common in market mechanisms such are economy conditions, management principles, and eco nomic units" (Laki, Szalai 2006).In the following lines, some of the research papers presenting general success factors are described.
A plethora of research is oriented on the transitory coun tries, in which the entrepreneurs faced to the completely new conditions and opportunities for the successful enter prise establishment on their own.Laki and Szalai (2006) examined key success skills of the postsocialistic Hungarian entrepreneurs.From his interviewbased research, the sociodemographic charac teristics dominated the success in the transiting Hungary."Age, gender, schooling, and residence mattered: in the en trepreneurial arena of the 1990s, it was a group of highly ed ucated middleaged urban dwellers that possessed a definite advantage ahead of all others" (Kaibori 2001).In addition to this, the employment history and general "luck" played the significant role as well.Kadocsa, Fransovics (2011) con ducted similar research among SME in Hungary, which was pointed to the impact of the EU accession on them.Main finding of this work was that "small businesses do not capi talize on the opportunities offered by the European Union, and do not make efforts to apply for EU grants and funds, or attempt to penetrate new markets" (Kaibori 2001).However, Kadocsa and Fransovics note that there are different fac tors that influence SME success such as cost management, trade and marketing, production, technical development and finances.Aidis, Mickiewicz (2006) conclude their research within the Lithuanian SME with the similar results as Laki and Szalai (2006) that the education of the entrepreneur plays important role in the success, but they find one more vari able, which turned to be more significant.They add the role of the export orientation as a key growth factor.Moreover, as a quite surprising fact could be that the "learningby doing" concept seemed to be not important factor of the transition growing companies.We must note that the nega tive influences, which might have impact on the company's success, are also required to complete the whole analysis.This paper was enriched for analysis of the negative influ ences like corruption factor which pointed up as one of the most influential.Among the negative effects, surprisingly, the Lithuanian enterprises did not comprise the taxes pay ments (Aidis, Mickiewicz 2006).Authors based this study on the concept of the multinomial logit estimator.
Jasra, Khan (2010) recognize the financial and techno logical resources, government support, marketing strate gies, and entrepreneur skills to have a significant and posi tive impact on business success among the 520 Pakistani SME (Jasra, Khan 2010).Within the researched companies, the study finds out that the financial resources are the key factor, which the enterprises selected as the most important.With regard to this, Jasra and Khan (2010) point out that this could be due to the poor governmental and banking support into the SME industry.The whole study was based on the responses from the questionnaire which were processed via basic multiple regression in IBM SPSS software.
The success factors of African SME are discussed in Monibo (2003), Harabi (2005), and Govindasamy (2010).First of all, Monibo (2003) introduces his work with a story of successful bakery in Nigeria which terminated its busi ness after the death of the entrepreneur.He states the main question if behind the whole success of a company lays only the founder's contribution.To accomplish this aim, the dynamic model techniques are used.In more detail, the survival function is depicted as the most suitable.To conclude this work, Monibo (2003) noted that succession problem of owner leave influenced almost 54 percent of companies significantly that they had to close within the next year.(Approximately, half of them, closed at once).Only 37 percent of the sample survived the "succession ex perience" (Monibo 2003).Harabi (2005), on the other side, explore the success factors of Macedonian companies as omitted variables from the growth equation of a company.This paper describes which of the factors contributes to the future size of an enterprise the most significantly.According to Harabi (2005), the important ones with positive effect are following: company location, diversification effect, le gal status, price competition, strong demand for product, governmental positive regulations.Negative factors, he recognizes the qualification of workers, small population centers, and governmental negative policies.As we can notice, the government appears on both sides.Last of the abovementioned papers is devoted to the South African minorities -Indians.In this work, mainly the causalities of differences between the successful enterprises were mea sured (managerial skills, personal factors, financing factor and ownership structure).These causalities were compared with the basic statistic values such as the mean, standard de viation, frequency and range.From the research, the educa tion factor and family involvement seemed to be as only the significant factors of influence.Number of years since the establishment appeared in the statistics but author rejected this variable because of lack of observations.Chittithaworn et al. (2011) suggests the factors which could help the entrepreneurs in Thailand to set up such a beginning standards or more likely investments into the company to get the growing and wealthy SME.
Collected 146 statistical responses showed that compa ny's characteristics, customer and market demand, coopera tion, marketing, external environment, and financial with other kinds of resources are significant enough to describe almost 60 percent of the model variations.Again, the mul tiple regression analysis is being used.Chittithaworn et al. (2011) find out that the Porter's Generic Strategies devel oped by Michael Porter generates conditions for successful startup SME growth in Thailand (cost leadership, market differentiation, product focus characteristics).
Research across the Taiwanese companies (Lin 1998) shows that on the first place, the entrepreneurs indicate the success in the progress in company's structure, tech nology, and development.After that, the managerial skills author deemed as much more significant than technical skills.However, the study concerned only 43 observations; it is one of the quite different approaches so far.
From the western background, a study on American familyowned companies by Motwani, Levenburg, Schwarz (2006), confirmed that most of family member's issues strictly relate to the enterprise performance, also "select ing a successor who possesses strong sales and marketing skills" is crucial for their success.One of the last conclusions was that the formal plan of succession and communication with the successor represents important role.The process of data collection author developed on the questionnaire survey among almost 4000 SME with 368 responses (which represents 9.2percentage response rate).Several limita tions of their model arise in the conclusion part, Motwani, Levenburg, Schwarz (2006) state that expect the lack of the available data their hypothesis could suffer from the subjective bias.It is because their analysis was based only on answers of CIOs and top managers.Moreover, the fact that this research was based on crosssectional industries could cause significant mistakes in final values for concrete industry study.Yusof, Aspinwall (2000) represent one of the industry specialized analysis, which was conducted on SME from the automotive sector in the United Kingdom.Their aim was to discover main success factors and total quality manage ment implementation.Response rate from the question naires was over 20 percent and the data analysis was done with IBM SPSS software.To verify the results, some of the tests on reliability were presented (consistency test, con struct and criterionrelated validity test).Concerning the results, among the English enterprises, the most significant factor seemed to be the management leadership abilities, results and performance reports, employee training, adopt ing a quality assurance system.On the other hand, the areas that actually need more attention from these companies are those having low practice levels, like continuous im provements of system, technical improvements, supplier quality assurance.Terminating remarks are devoted to the recommendation.We could perform similar analysis on other industries to compare the crosssectional differences in SME or as well to introduce the extension of timebased performance of the companies.
Most recent research on the SME success factor analy sis is based on the enterprises at the time of EU Accession (see Kadocsa, Fransovics 2011).Rural enterprises did not find the benefits coming from the possibility of applying for grants or funds in the EU, but on the other side, they surprisingly evaluate as an important factor the age (prob ably the experience) of the manager -founder of the SME.In addition, there appears to be a correlation of the micro and macroeconomic conditions between the poor and rich areas.
Overall, it can be stated that the analysis of SME suc cess is crucial from various reasons, starting with the fact that government policies have the substantial influence on them (Harabi 2002), and from other reasons like banking success scoring, consultancy services analysis (Peacock 2000).Researchers are usually consistent in their com ments on indicators, or more rather factors that influence the SME success rate positively or also negatively.Most of them agree on the education factor, employee training (Laki, Szalai 2006;Aidis, Mickiewicz 2006;Peacock 2000).But other influences are obvious from the researches: age of manager -owner (Laki, Szalai 2006); market skills and experience of the selected manager (Laki, Szalai 2006;Yusof, Aspinwall 2000;Monibo 2003;Motwani et al. 2006); ownership structure (Hanousek et al. 2005;Hanousek et al. 2012;Estrin et al. 2009); product orientation and trade (Aidis, Mickiewicz 2006;Jasra, Khan 2010;Chittithaworn et al. 2011;Porter 1980); government influence (Kaibori 2001;Aidis, Mickiewicz 2006;Jasra, Khan 2010); and many others.
Importance of ICT sector analysis is underlined by Kaibori (2001) with hightech companies in the Silicon Valley in the U.S. Czech ICT sector has gone through a "revolutionary" period in last two decades, and yet, it still has a potential to develop.Thus, we should consider the analysis of different factors influencing their success as im portant and valueadding topic of research.

Data and methodology
Our analytical part is heavily based on the questionnaire survey distributed among the Czech ICT companies.Information about individual enterprises including contact emails was obtained from ČEKIA Magnusweb database.Our selection of SMEs corresponds to the CSO selection of the ICT sector with 39,463 records in total.All companies are selected with regards to filled email contact address from the ICT sphere which represented 7,979 potential respondents.Email was messaged to all contacts with a request to fill the survey.
The survey was active for approximately one month, during April, and the respondents were asked in one wave overall.After this round, only 47 answers were collected, which could not be considered as a sufficient or reliable sample.Second round followed after, increased the number of completed questionnaires to 131.
Questionnaire distributed among the ICT companies consisted of three parts including thirty direct questions and three optional supplement questions to provide an oppor tunity for respondents to express their ideas and opinions.Various types of questions were posed -multiple choices, yes/no, pick one and rating, and description questions.
Most of the respondents were companies from the soft ware development field (56.19 percent), into this category be longs enterprises, which offers products of the nonmaterial character -website, software development.Second largest group represent enterprises that distribute the software so lutions.These are the resellers of software and companies, which implement the software solutions (20.95 percent).Two smallest groups are analogically the hardware distribu tors (17.14 percent) and hardware developers (5.71 percent).
Companies, whose fields of activity extend more than one category, were added into more categories at once.
Figure 3 above hence denotes that most of the companies in Czech are oriented in the field of software development.This might be logically caused by the fact that it is still easier to create small business company, which develops mobile applications or internet pages than to establish a factory de veloping new components or devices, moreover in the time of cheap and easy supply from Asian countries.According to some of the commentaries, this hypothesis is confirmed later in this chapter.
These companies are in the most cases owned by one native legal personality (38.1 percent of observations) or one legal entity (25.71 percent).Third largest group repre sents the group of Czech owners (19.05 percent) followed by the groups of Czech legal entities (15.24 percent), which demonstrate that the ICT sector in Czech Republic still does not attract foreign small investors.Logically, than the small est group of proprietors are the foreign owners.From the general structure of the sector, it could be assumed that companies, which have foreign owners background, do not come under the group of SME.On average, companies employ 11.94 employees, where 2.05 are on the managerial post.According to Gupta (2010) span of control range is generally quite wide.It represents the 4-22 employees under the managers lead, when the number depends mainly on the nature of the work.Moreover, companies with more traditional orientation believe that 5-6 subordinates per supervisor would be ideal, and more innovatively based companies prefer the rate of 15-20.Nevertheless, if the nature of work cannot be precisely determined from this research (mainly hierarchy of the individual enterprises), we could state that the result mostly fits for this division.
Common ICT SME in Czech Republic operates for nearly 13 years and there is 40 percent change that it has a headquarters in the capital city.This statement is based on the fact that average age of a company in the sample is 12.96 years and exactly 40 percent of them operate in the city with more than 1,000,000 of inhabitants, which cor responds only to one city in Czech.From the company age distribution (Appendix 5) is noticeable that there were two periods of establishment that could be quite obvious.Older group of enterprises are around the 20 years old.This means that they were settled just at the beginning of 90' , when the market was opened to any new companies and it of fered many opportunities of focus.Second group represents younger SME than 10 years.This wave might be supported by the increasing trend information technologies in Czech Republic.Amount of internet users increased, web pages presentations boosted, mobile devices started to be more used with wider possibilities of usage, and the new drift of data analysis occurred.All these are the trends, which became popular around 10 years ago and they created the new spot on the ICT market.This statement is supported by the fact that 52.54 percent of analyzed enterprises from software development industry begun their activities after 2003 and Voříšek (2010) states that the ICT services sector tend to be more progressive than the manufacturing sector.
Concerning the geographical distribution, the largest group of respondents actively operates in the capital city (40 percent).Second were surprisingly the cities with less than 50,000 in habitants (the smallest offered category) with the share of 32.38 percent.Next two categories includes cities with 50,000 to 500,000 inhabitants where jointly operates 23.81 percent of companies.Smallest group belongs to 500,000-1,000,000, where only 3.81 percent of enterprises operate.
Frequently discussed topic that influences the company success is the age and the educational level of the CEO of the firm.From the available sample, managing directors are most often between 35-45 years old with finished first degree academic education, which could be expected based on the empirical studies of employee productivity.Joint dis tribution showed that the most common situation is when the CEO is between 31-45 years with finished high educa tion or CEO between 36-55 years with finished firstdegree academic education.Separate percentage distributions of age and education are listed in the tables below.This analy sis has just the informative character, closer description of connection of CEO age and education and understanding of success is mentioned in following paragraphs.
Next point of view is focused on the markets that the ICT firms operate on.The largest portion belongs to the do mestic market (46.96percent), followed by the regional do mestic market (38.46 percent).Unexpectedly, the smallest parts represent the foreign markets.EU market is impacted by 11.72 percent of analyzed enterprises and there are only 2.86 percent of globally active companies.Thus, it could be implied, that enterprises do not tend penetrate to foreign markets, which might limit their possibilities and capital inflow from nondomestic markets.Moreover, this finding looks surprisingly, because of the fact that in previous para graphs it was mentioned that the largest representation in the sample is by the software distributing companies, which do not usually have physical boundaries of sale.
Furthermore, one the finding about most frequent the channels of sales underlines the missing possibility of spreading into foreign markets.Enterprises from 53.33 per cent offer their services and products via internet that could possibly open the barriers of their expansion.
Nevertheless, second most popular method is by direct contact with clients (36.97 percent) and remaining share belongs to retail and wholesale distribution.Comparison that is more detailed is stated in the figure above.In addi tion to this, one of the companies had a commentary on different way of sale and that is the foreign sales distributor.
Another topic concerns the company's clients and di rect competition on the market.Because there is quite large variation between the collected amounts, the exact num bers would not have any explanatory value.However, the first differences marks of the observed years would reveal the trends in growth or loss of the clients.From this point of view, majority of the analyzed companies in 2011 reg istered increase in the quantity of clients (45.71 percent).Last year, even 49.52 percent of enterprises noticed growth of their customers.This was projected to the companies, whose customers stayed the same between 2010 and 2011 (31.43 percent).For 2012, it dropped to 22.86 percent of firms.The amount of enterprises, which lose customers, is increasing during the observed period.(The growth was from 22.86 percent between 2010 and 2011 to 27.62 percent between 2011 and 2012.)For the current situation in 2013, alternatively, most of the companies expect increase of their customers by more than 5 percent (35.24 percent).The same numbers as last year expect 32.38 percent of enterprises and 19.05 percent expect slight growth.The rest of them (13.33 percent) do expect diminution.
Financial resources represents quite important compo nent of the success of an enterprise.Several questions arise with this topic.Can the company finance all its expenses by the revenues from previous years?Does it need bank loans or the insufficient funds comes from the owners invest ments?Do enterprises use available financial subsidies for example like EU grants?This paragraph analyzes collected answers about the financial resource differentiation between the ICT SME.Generally, most of the firms are independent from others subsidies than their previous earnings from sales.All of the categories are around 40 percent selfreliant.Secondly, domestic investments represent alternative way of supporting the operations.They are used to support mostly in case of HW distribution enterprises.Hardware develop ing firms more likely support their operations banks loans compared to other ICT SME categories.This might be due to the fact that it is quite difficult to establish stable hardware developing company on the Czech market, maybe because of the competition of large corporations, as some of the respondents noted in the commentaries part.
Software companies are only users of financial subsidies and grants.Distributors use them in 13.79 percent of cases.Still they are able to cover their expenses in most cases by sales earnings.Summary of this analysis is stated in the Table 1.
To conclude the success identification of companies, the analyzed companies choose as the most informative unit of success measurement the financial factor earnings where the yearly revenues and amount of customers play important role.Moreover, they identify the key factor the originality and accessibility of their services and products as crucial to have competitive advantage.The predictive value of abovementioned indicators is discussed in the following chapter.

Empirical model and its testing
By determining the predictive value of different variables for success rate factors, companies will be able to enhance their internal operations to gain better performance and results.Several other reasons why these processes are ap propriate to analyze are summarized in the final parts of this paper.Econometric models could perform such an analysis.
Process of identification of individual success factors determinants includes all variables into statistic software that stipulate their significance in separate cases.Then, by removing less significant ones, which do not have any pre dictive value, the final model of given dependent variable is constructed.Results are then discussed and controlled.The outcome should be that they have overall meaning for the purposes of the paper and that they explain some rational ef fect.Even though, most of the responses are complete, some of the observations are missing in the data set because of the fact that the financial values question was optional and some respondents skipped it.These are usually continuous vari ables.Other type of responses, discrete ones, are ordinarily complete, thus if the explanatory variables are compounded only by them, amount of observations is complete for this case.The most common methodology of this research is generally usage of linear regression.Thereafter, the models are based on the linear relationship between variables Y i and X i in the form: where Y i denotes the ith observation on the dependent variable Y which represent success rate indicator, X i rep resents ith observation on the vector of independent, ex planatory variables, n is number of collected observations, and α and β are intercept and slope of the simple linear relationship between Y and X.
In order to get the predictive value of this model, it is important to estimate the intercept and all included co efficients.This step is done through ordinary least square method, where the sum of squares of error terms in the regression is being minimized.
All the calculations are processed in the IBM SPSS Modeler software, which works on the basis analyzing and performing individual series of nod steps that comes into the final model outcome.
Results of individual regression model are applied in order to identify the factors that are most significant for enterprises success.Opening part of each interpretation stats with the model notation, where the bold numbers state for coefficients values followed by the robust standard errors value in the parenthesis and *, **, *** denoting the signifi cance level less than 1 percent, 5 percent and 10 percent.The results of the stepwise regression estimates are presented in Table 2.
It appears that zisk_zam_SL on which the Q72010 vysledek_hospodareni_SL, logarithm of earnings in 2010 has positive influence with the dummy variable if the com pany's main financial resources are from domestic invest ments (Q9moznost1=1) or from earnings from previous years (Q9moznost1=3).Dummy variable is recognizable by the "="character behind the variable name and it stated on which of the possibilities it is applied.On the other side, negative influence has an expected growth (Q82013 zamestnanci=2) or stay (Q82013zamestnanci=3) in the costs on employee training in actual year.In addition to this, negatively appear also if company finances its opera tions from the bank loans (Q9moznost1=4).Insignificant value, also proving problems with colinearity, is devoted to the number of customers in years 2010 (Q242010) and 2011 (Q242011).
Overall accuracy of the model represents 55.7 percent, where the information criterion 0.599 labels this model as quite stable, but with possible irregularities.
The next model indicates company's success based on the average revenues performance in past three years (2010)(2011)(2012).This variable was reported as the most important for the respondents in the previous data analysis.Model selec tion method was performed by the best subsets procedure.In addition to the data set, the automatic data preparation technique was applied.By performing this technique, either the nonsuitable and outperforming observations are omit ted or the categories of some discrete variables are merged to maximize association with the target (predicted) vari able.The results of the stepwise regression estimates are presented in Table 3.
Here, trzby_SL_prumer are computed as average log arithms of total revenues for last three years.Q72012 naklady_transformed represents the positive added value of the total costs in last observed year.The intercept of this variable is too small because the logarithm transforma tion was not applied on it.Other variables with positive value are not such significant in this model, these are the Note: *, **, *** denoting the significance level accordingly less than 1 percent, 5 percent, and 10 percent.
Business: Theory and Practice, 2015, 16(3): 304-315 Q29nakladynazamestnance_transformed (companies for which the costs on employees present a barrier in their development) and Q13_transformed (age of a company).The fact that contributes negatively is the companies which do not adapt their revenues as important part of operations (they do not gain much gravity to them -Q12obrat_transformed=0, Q12obrat_transformed=2).The positive value of the middle variable might be caused by lack of available observations, but in this model, it implicate opposite effect than its border dummy variables.In addi tion, one of the significant elements is the age of company CEO.Those enterprises, whose managing director's age is between 18-30 years, perform with lower average returns.Surprisingly, if enterprise is planning to invest more into marketing in following year (Q82014marketing_trans formed=0), it also indicates worse results on returns.
Summary statistics of this econometric model demon strate better results than the previous one.Total accuracy as it is illustrated below reaches 72.5 percent and the informa tion criterion -70.129 indicates well fitting model.
Moreover, the normal distribution of residuals seems fitting except a small rigidity on the left tail of the distribu tion and the colinearity test did not showed any problems with the variables.
In order to enlarge the idea of using revenues as a pre dictive function of success of individual companies, if the revenues from last observed period are taken, some signifi cant explanatory variables are found as well.The results of stepwise regression (method of best subsets selection) are presented in Table 4.
In this case, the logarithm revenues in observed pe riod (Q72012trzby_SL) are logically positively and most significantly dependent on logarithm of total costs in the same period (Q72012naklady_SL), on the logarithm of marketing costs in 2010 (Q72010marketing_SL) and the amount of employees (Q5).Questionable could be that the marketing costs from previous or observed year did not oc cured as significant.This might be due to lack of collected data.Aleternatively, only negative coefficient are proved for logarithm of total costs in previous year variable (Q7 2011naklady_SL).
Overall, this model perform better than the previous again.Almost the same effect is explained by less variables which have obvious influence on its value.Thus, the most important predictor in this is represented by the total costs of acutal year.Others do not reflect so much importancy.Even though the information criterion reaches -81.779, normality distribution comparison with the residuals do not show much comparable course.Colinearity test confirm above mentioned hypotheses for both total costs.Thus the previous model which predicted revenues should be pref fered for these conditions.
Testing R&D values showed that for both possible approaches (graveness approach and continuous values model) are possible some significant predictors.This is the reason why in this part are discussed both of them.The results of estimations for R&D importance model (forward stepwise regression model best subsets selection method) are presented in Table 5.
Here the importance (success) value of the research and development (Q22rad) is characterized by positive affili ation to various types of ICT companies.The most impor tantly the companies from hardware development sphere (Q3=2).Other specifications contribute to this success fac tor with smaller proportion software developing (Q3=1) and software distributing (Q3=3) companies.Hardware distribution does not show any significance important for this model.Positively contribute also if the company consid ers the earnings as an important part of company perfor mance (Q12rozdelitelnyzisk) and logically most important is how much they spend on R&D (Q72012RaDSL) and if they consider the active investments into development as a comparative advantage (Q12rad).Negative coefficient represents the logarithm of investments into R&D in 2010 (Q72010RaD_SL).
Total accuracy test shows 58.5 percent accuracy.where the information criterion is heavily high valued by rank 104.688.which indicates that the model does not need to fill well.Distribution of residuals on other side quietly well describe the normal distribution and in addition the impor tance of individual predictors is equally divided.And the colinearity problem is not spotted among these indicators.Secondly, seemingly better model represents the R&D investment in observed year 2012.By using the best subset method of estimating the regression, the results are pre sented in Table 6.
The amount of R&D invesments (logarithm) in observed year (Q72012RaD_SL) depends mainly on positive finan cial indicators of logarithm of R&D investments in previ ous year (Q72011RaD_SL).and logarithm of total costs in actual (Q72012naklady_SL).Neglectible positive con tributor is also the barrier of lack of employee qualification (Q29kvalifikace).Q72011naklady_SL.previous year total costs are represents strong negative influence on the model in combination with the less significant predictor of invest ments into R&D in 2010 (Q72010RaD_SL).Logarithm of investments into marketing in 2010 (Q72010marketing_SL) also negatively inflence R&D investments this year.
Although the overall statistics predicts a very stable model with information criterion being very low (-259.324)and the distribution of residuals quite normal, colinearity in this case is quite high almost for all financial variables that it might lead to inaccurate numerical results.
As it was mentioned above, several problems arose dur ing the construction of individual models that cannot be enhanced with available data set.The crucial represents insufficient amount of observations among ICT SME.This could be corrected by using more extensive long lasting research.In addition to this, with wider possibilities of re search would be connected a collection of more variables, which were marked by companies as important ones in previous chapter.Thus, more appropriate and quantifiable results would be achieved.Finally, if there would be such a variables, complex econometric model would be possible to apply to gain better described individual effects that would not indicate any statistical problems.
The constructed models above suggested that there is a negative correlation between company performance and the Business: Theory and Practice, 2015, 16(3): 304-315 lack of skilled employees.From this fact, it can be implied that freshly educated students who have the most upto date knowledge of modern trends are one of the factors that positively influence a company's success.In addition to this, young people without longterm experience in the labour market represent one of the cheaper working forces available, thus their employment and induction into en terprise processes might include lower costs on the side of wage expenditures and, simultaneously, motivate employees who are willing to grow in knowledge and performance.Unfortunately, this paper did not contain statistics about the movement of this labour force.
Neither the political situation, nor fees or taxes occur in any of the models significantly.Even from the descriptive statistics section, companies do not place much importance in the effect of political situations on the success ratio.
It can be gathered from the statistics illustration in the ICT sphere development description that the financial crisis of the previous years has hit this sector markedly.This ef fect probably carried over into the modeling part and also can be seen through the lagged influence of several types of investments (a frequently used variable was the marketing costs from the previous two years).

Conclusions and policy implications
Overall it becomes obvious that Czech SMEs, similar to other SMEs in small open economies, face difficulties in achieving success within their respective markets.Some of them dem onstrate more sufficient results than others; others are satisfied with their profitability at a specific return level.Conversely, there are some that need organizational help to cope with their position.These variations are based on the indicators and factors that have a connection with their success rate.
Our research indicated both financial and nonfinancial indicators such as earnings and revenue value, divisible prof its, number of customers, company publicity, active R&D, and participation on large projects.This identification led to the possibility of modeling the success rate among the sample of answers.The ratio of earnings per employee as the first indi cators of a company's success was chosen from the studied literature.Secondly, based on results from the descriptive section, the average returns value is quite significant.Finally, the gravity of R&D is the last model to have been constructed.
The first model logically depends, for the most part, on the historical value of earnings and expected decline in employee expenditures.The upside for this success rate is that the company is able to cover its expenditures on opera tions from either previous sales or from domestic invest ments.On the downside, bank loans discriminate against the success valuation.
The second model is based on the total costs in an ob served period.If an enterprise does not regard revenues as important, it has a negative influence on the model.If not, it is considered to be an unintended effect.Similarly, young directing managers do not contribute to revenue perfor mance positively.
The last prediction concerning the magnitude of re search and development in a company is determined by which ICT sphere said company operates in.The most favored companies are from the hardwaredevelopment sphere.It is also dependent on historical R&D investments and divisible earnings.
The rationale of individual models is than discussed in the result interpreting section, where the proper specifica tions of answers on the stated hypotheses are presented.Overall, our results from this analysis could be used in con sultancy as well as in banking.Those institutions would then be able to identify enterprises which tend to perform better on the Czech market.Via such analysis they would be able to maintain a rationalesupported attitude in different situations.In addition to this, models on the success rate of ICT companies would be appropriate to be used by the government because these processes might enhance pos sibilities of financing individual projects and spheres.All the abovementioned models could then serve as starting points for wider research among ICT companies to be enhanced by a wider range of observations and thus help this recently stagnating sector to show its real importance.

Fig. 1 .
Fig. 1.ICT sector employment development.Number of people employed in ICT sector in thousands of people and employment in ICT as percentage of total employment in all sectors.

Fig. 3 .
Fig. 3. Field of activities distribution of ICT SME (created by the authors)

Fig. 4 .
Fig. 4. ICT SME market orientation and channels of sale distribution (created by the authors)

Table 1 .
Financial structure of Czech SMEs in ICT sector

Table 2 .
Results of the earnings per employee model.

Table 3 .
Results of the stepwise regression estimates (best subsets method of selection).Company's success model.Dependent variable is average logarithms of total revenues for 2010-2012

Table 4 .
Results of the stepwise regression estimates (best subsets method of selection).Company's success model.Dependent variable is logarithm revenues 2012 Note: *, **, *** denoting the significance level less than 1 percent, 5 percent and 10 percent.

Table 5 .
Results of the forward stepwise regression estimates (best subsets method of selection).R&D importance model.De pendent variable is importance of active R&D for success