«VOLUM E 1 1, N UM B E R 1 I S SN 2 1 6 8 - 0 6 1 2 F L ASH DR I V E I S SN 1 9 4 1 - 9 5 8 9 ON L I N E T h e In s t it ut e f o r Bu s i n e s s an ...»
As a complementary technique to the discriminative analysis, the logistic regression is used, expecting that this regression will help to validate the previous results and will offer greater information. Using these methods, the information will be detailed but without being redundant. Because the proposed model will be the most complete and will allow differentiating between innovating companies from which are not.
The database used in this analysis is the Survey on Research and Development, Technology, Linking and Innovation in Mexico, made by the Consejo de Ciencia y Tecnologia of Mexico (CONACYT) and the Instituto Nacional de Estadistica, Geografia e Informatica (INEGI). The objective of this survey was to collect data of the resources, the activities, the attitudes and the results associated with the generation, the acquisition and the development of technology in Mexico. The inquired population was constituted by the establishments and the institutions of the public administration sectors, superior education, organizations without profit aims and the productive sector, in particular the manufacturing one, due to the fact that this sector is the most dynamic in technology incorporation within the machinery and the equipment.
The statistic formulation of the sampling is probabilistic so that the data can be extended to the population.
The population was obtained from lists of the Economic Census of 2010, directories of firms affiliated to entrepreneurial organizations such as Camara de la Industria de la Transformacion (CANACINTRA), GCBF ♦ Vol. 11 ♦ No. 1 ♦ 2016 ♦ ISSN 1941-9589 ONLINE & ISSN 2168-0612 USB Flash Drive 232 Global Conference on Business and Finance Proceedings ♦ Volume 11 ♦ Number 1 CONACYT and the Secretary of Economic Development in Mexico. A simple stratified sampling was made by size of company and sector of economic activity. The sample was selected independently for each sector. The establishments entered with certainty were the ones with 250 employees or who had received allowance from CONACYT in Mexico for research projects in cooperation with some university or public investigation center. The multipurpose orientation of the survey forces to make a sampling sufficiently big to calculate all the variables of interest, so that the size was calculated based in the main variable: the spending in R&D.
The calculated size differs in the cases of establishments that during the data collection were not situated, they were in strike or were broken or suspended its operations. With an answer rate of 90,3% 398 questionnaires were answered. The sector distribution of the sample is represented in Table 1. The data were collected from June to September of 2010. First the questionnaires were delivered to the establishments for their answer, the also received a guide with instructions. In a second visit the questionnaires were collected and, just in case, the informers were helped to answer it correctly. The capture and tabulation of the data was made by the INEGI, this institution gave an analyzed database to the CONACYT. The information was processed with the statistic software SPSS 10/12.0. For the managing of
the data the following models were established:
Discriminative Analysis to examine the characteristics that differentiate an innovating company from a non-innovating one and Analysis of logistic regression, to support the previous one.
RESULTSDescriptive Analisis When using the subjective definition of innovating behavior, 26,6% of the companies described themselves like innovators; whereas with the objective definition the percentage happens to be 55.3%. These numbers show that firms underestimate their activities associated with innovating behavior. Discriminative Analysis is a multivariable quantitative method, which takes a qualitative variable as dependent; and one or more quantitative variables as independent ones. The results are lineal functions called discriminants, which allow classifying the individuals in one of the categories established by the values of the dependent variable.
In this case, from a set of variables that tries to gather the characteristics of the company (structure, behavior and performance) the companies are classified in innovating and non-innovating companies. Two discriminating analysis were made, one takes the innovative behavior as the dependent variable according to the subjective answer of the surveyed company and another that defines the groups based on proposed objective criteria. The rate of classification found with the first criteria (subjective nature) was 75.4%, while with the second criteria (objective nature), was 77.1%. These results confirm that the innovative behavior objectively defined improves the empiric classification, this is, increases the rate of correct answers when the desire is to distinguish an innovative company from other that is not. The quantitative variables with major discriminative power are both the percentage of sales dedicated to machinery purchase,
as well as, the percentage of technology acquired abroad. The proposed discriminant model is:
D= 0.035 (OCDE) – 0.26 (Maq99) -0.12 (Tec_Ex99) + 0.658 (Camb_Org) + 1159 (ID_Ing) + 0.728(Apoyopub) - 3918
When this equation is applied to the centroids of the groups several negative values are obtained for the innovative companies and positives for the non innovative companies. So, the negatives quantitative variables are associated with the innovative behavior. By means of this discriminative equation it is observed that for each unit that increases the percentage of sales dedicated to the purchase of machinery and equipment, the discriminative function (equation) decreases 0.26. When the percentage of the GCBF ♦ Vol. 11 ♦ No. 1 ♦ 2016 ♦ ISSN 1941-9589 ONLINE & ISSN 2168-0612 USB Flash Drive 233 Global Conference on Business and Finance Proceedings ♦ Volume 11 ♦ Number 1 technology purchased abroad modifies by a unit, the discriminative function will do it negatively in 0.12.
So, the bigger the effort to acquire technology of a company and as the firm increases its technological stock from abroad, there will be more probability the company is in fact innovative.
With regard to the rest of the variables (Camb_Org, ID_Ing and Apoyopub), the bigger values they take, the more possible will be that the company is in fact non innovative. It is worth saying that the way the codification was made, influences the interpretation. The value of “one” was assigned to the presence of the attribute and the value of “two” was assigned to the absence. So, the top values refer to the absence of the attribute: lack of organizational change, lack of formal units of R&D and engineering, lack of government aid. The absence of this attributes is associated with the non innovation. The activity sector (OCDE) is a nominal variable with values that do not have an ordinal logic, so is not useful in the equation to get the truly membership to a group. This variable, along with others of qualitative type, will be also explored in the logistical analysis to prove its influence. These results allow rejecting a group of variables that were though that could have any discriminative power among the studied companies; between them
the following ones are emphasized: Of structural type:
Size (sales as well as employees) Foreign participation in the equity of the firm.
Participation in foreign markets.
Grade of specialization in its products, and Age of the company.
Of technology behavior:
Number of engineers working in the company.
Years of Experience in the engineering department.
Of competitive performance:
Growth Rate of sales.
Growth Rate of hiring.
With this analysis is possible to classify 77.1% of the total number of companies, of which 75.5% are innovative and 79.2% are not. In the Table 2, it is also shown the results of the classification made with the technical of cross validation, this is, one case is left outside and every case is classified using the discriminative functions derived from all of the other cases.
The logistic regression is a multivariable technique with the objective of classifying the individuals in one of the two groups established by the dependent variable. Once it is known the model, an individual might be classified, in a probabilistic form, in one of the answer groups when the value of the explicative variables is known, but not the kind of them. In this case, according to the logistic regression the variables associated
with the innovative behavior are, by associative power:
Acquisition of machinery.
The existence of a formal unit of engineering or development.
The possibility of public aid.
The introduction of changes in the organization of work.
The acquisition of machinery and equipment abroad.
The subsidiary nature of the company.
GCBF ♦ Vol. 11 ♦ No. 1 ♦ 2016 ♦ ISSN 1941-9589 ONLINE & ISSN 2168-0612 USB Flash Drive 234 Global Conference on Business and Finance Proceedings ♦ Volume 11 ♦ Number 1 Y= 0.089 + 5837 (Adquima) + 4356 (ID_Ing) + 2428 (Apoyopub) + 2291 (Camb_Org) + 1959 (TIAE) + 2.76 (Subsidia)
Adquima is the condition of having acquired or not machinery and equipment between 1999 and 2002.
Camb_Org is the condition of having done an organizational change.
ID_Ing is the condition of having a formal unit of investigation and development or engineering.
Apoyopub is the condition of having received any public aid.
TIAE is the condition of having acquired technology in machinery or equipment abroad.
Subsidia is the condition of been or not a subsidiary company.
The previous equation can be interpreted using an example. If two companies with similar characteristics, practically equal, in its position with respect to the purchase of machinery, organizational change and public support, the fact that one had a department to the investigation and the development or enginery and other no, multiply by 4 356 the probability that the first company was innovative. Then the fact that a company has these attributes increases the probability of which it presents an innovating behavior. On the other hand, the variables that cannot be used as a criteria to explain the innovating behavior between the studied companies are localization, the economic activity sector, the size, the foreign capital, the diversification of the product portfolio, the rank of the amount of sales, the technological level of the economical sector, the use of e-mail, the source of the machinery and equipment supplier (intra-enterprise and national), and the origin of the source of provision of unreal goods and the expansion of the employment and the sales.
From the analyzed models and variables of the production system in Mexico, it is concluded that firms tend to underestimate the activities and the results that can be associated to an innovating behavior. In particular, the condition of having and employing complex machinery and equipment is not valued; as a result an infra-valuation emerges. Neither the structural characteristics nor the competitive performance help to explain the innovating behavior of the studied companies. Evidently, the innovators distinguish themselves for having formalized their activities of engineering or R&D, having conducted changes in work organization during period 2008-2010 and having obtained some governmental support. The previous claim had been verified empirically with two tools of multivariant analysis. The main activity associated to innovation is R&D. As many authors  have suggested, this is not the only source of innovation, but this relationship seems irreplaceable due to the necessity of new technology incorporation. On the other hand, the technological learning has been associated to the production routine  that is why to have formalized the activity of engineering facilitates the codification of the knowledge; when providing an organizational support to the accumulation of technological capital in the companies. This favors the incorporation of new knowledge and solutions, generated in the solution of problems derived from the productive task and the development of technical abilities that reduce the difficulties attaché to the transference, adoption and adaptation of new technology.
The exigencies of the new competitive environment, in particular the implantation of transnational companies in the national markets and the necessity to participate direct or indirectly in networks of global production have pressured intensively. In response the companies have carried out organizational changes that allow them increase the quality and the quantity of their production, as to adopt shorter delivery times.
This condition of fast adaptation to changes in turbulent surroundings has been called adjustment or fitness. Then, since the companies that had made organizational changes have, more probability of being GCBF ♦ Vol. 11 ♦ No. 1 ♦ 2016 ♦ ISSN 1941-9589 ONLINE & ISSN 2168-0612 USB Flash Drive 235 Global Conference on Business and Finance Proceedings ♦ Volume 11 ♦ Number 1 innovating, we can say that those that present capacity from adjustment to the surroundings can be grouped in the group of the innovators.
In the survey, there are not indicators that measures techno-productive and social relations in the companies.