Analyzing the Medium-Low and Low- Technology Firms’ Innovative Behavior in an Emerging Economy

Análisis del comportamiento innovador de las empresas de tecnología media- baja y baja en una economía emergente

Análise do comportamento inovador de empresas de média-baixa e baixa tecnologia em uma economia emergente

Esta obra está bajo una Licencia Creative Commons Atribución-NoComercial-SinDerivar 4.0 Internacional.

Innovation has caught the attention of academia, governments and business managers alike. Academics would like to know what motivates companies to innovate. Governments seek to foment innovation because, according to Ahlstrom (2010), innovative firms generate economic growth and employment. Business managers are interested in innovation because it allows them to generate competitive advantages (Urbancova, 2013) and improve the performance of their companies (Jansen, Van Den Bosch and Volberda, 2006).

In the literature on innovation, most of the studies draw their data from developed countries and research high-intensity technology firms (Hervas-Oliver, Garrigos and Gil-Pechuan, 2011). The relationships between absorptive capacity and both technological innovation (Ali and Park, 2016) and organizational innovation (Chen and Chang, 2012) have also been analyzed. How organizational innovation mediates the relationship between absorptive capacity and technological innovation (Camisón and Villar-López, 2014) and how the acquisition of machinery, hardware and software improves innovation capability (Santamaría, Nieto and Barge-Gil, 2009) have been studied as well.

The first contribution of this study is that it analyzes the innovative behavior of manufacturing firms in Peru, an emerging Latin American economy. As Latin American economies face the double challenge of needing to keep growing while at the same time reducing levels of poverty, understanding and evaluating the way that Latin American firms develop innovation capacities is critical (Olavarrieta and Villena, 2014). A second contribution is that this study focuses its attention on the relationship between non-technological and technological innovation. Most studies have analyzed how organizational innovation is related to technological innovation (Camisón and Villar-López, 2014), but these tend not to take into account marketing innovation, which is one of the key elements of non-technological innovation.

The third contribution is methodological: since most of the aforementioned studies are cross- sectional studies; for this research, a repeated

The third contribution is methodological: since most of the aforementioned studies are cross- sectional studies; for this research, a repeated

cross-sectional design was applied using the database of two national surveys of innovation in Peruvian industry corresponding to the years 2012 and 2015. Therefore, it was possible to measure invariance, thus providing an opportunity to verify that the averages and compound variances were equal in the two groups. The groups were then compared to identify the change in the innovative behavior of medium-low- and low-technology firms (Mathews, 2017). This approach enabled a response to the research question: How did Peruvian manufacturing companies change their innovative behavior between 2012 and 2015?

Based on this question, the approach aims to explain five relationships, namely: (i) absorptive capacity and technological innovation; (ii) absorptive capacity and non-technological innovation; (iii) non-technological innovation and technological innovation; (iv) the acquisition of machinery, hardware and software and technological innovation; and (v) how non-technological innovation mediates the relationship between absorptive capacity and technological innovation.

It is worth noting that the context of this research is the Peruvian economy, which has shown sustained growth (Scott and Chaston, 2012) making it one of the fastest-growing economies in the region before the commodities crisis in 2014 (Brenes et al., 2016), which forced companies to face a reality with the following characteristics:

(i) a government that promotes open innovation (Ramírez and García-Peñalvo, 2018) and exports (Salas and Deng, 2017); (ii) companies that invest very little in research and development and prefer to innovate by buying machinery, hardware and software (Tello, 2017); (iii) companies that face informal competition (Heredia et al., 2017) and have problems obtaining financial resources to promote innovation (Pérez et al., 2018).

The unit of analysis is Peruvian manufacturing companies that participated in the national innovation surveys of the manufacturing industry in the years 2012 and 2015 and that presented a medium-low and low-technological intensity.

The structure of the present study is as follows: first, the theoretical framework and the hypotheses are presented; second, the methodology is shown; third, the results are given; and finally, the

The structure of the present study is as follows: first, the theoretical framework and the hypotheses are presented; second, the methodology is shown; third, the results are given; and finally, the

discussion, conclusions, limitations and suggestions for future research are presented.

This theoretical background begins by defining the key characteristics of medium-low and low-technology firms, absorptive capacity, and acquisition by medium-low and low-technology firms and proceeds to generate the study’s hypotheses for the five relationships outlined above.

As shown by the fourth European Community Innovation Survey (CIS4), which analyzes medium- low and low-technology firms, these two firm types tend to be characterized by innovations in processes, organization or marketing and have a high dependence on an external supply of technologies in the form of machinery, hardware and software (Heidenreich, 2009). In these firms, the role of formal and informal knowledge is important, as it has been discovered that they innovate beyond activities directly related to research and development (Sciascia et al., 2014; Santamaría, Nieto and Barge-Gil, 2009).

Absorptive capacity (AC) is the firm’s ability to recognize the value of new, external information in order to assimilate it and apply it for commercial purposes (Cohen and Levinthal, 1990). AC has had a major impact on organizational research and has attracted the attention of a large number of researchers, as it is the capability that most influences the development of competitive advantages and firm performance (Volberda, Foss and Lyles, 2010).

Arbussa and Coenders (2007) contend that the acquisition of machinery, equipment and hardware is one of the activities carried out by firms to improve their innovation capability. Moreover, Frank et al. (2016) researched innovation in Brazil and pointed out that the purchase of machinery and equipment had a positive effect on the innovation capabilities of those firms.

The present study is based on data obtained from two different waves of a national innovation survey of the Peruvian manufacturing industry carried out in 2012 and again in 2015. The Instituto Nacional de Estadísticas e Informática —INEI— collected the data. INEI surveyed Peruvian manufacturing firms using a questionnaire developed according to the Bogotá Manual, which is based on the Oslo Manual.

INEI conducted the first Peruvian innovation

survey in 2012, collecting information for the period

2009-2011 from a representative sample of 1220 large, medium and small firms from different Peruvian regions. From this database, 856 medium- low and low-technology firms were selected for the present research. Meanwhile, from the 2015 innovation survey, the information gathered belongs to the period 2012-2014 and consisted of a representative sample of 1452 large, medium and small firms from different regions. From this database, 1106 medium-low and low-technology firms were considered for this study.

Figure 1 shows the conceptual model of the relationship between the four constructs: technological innovation, non-technological innovation, absorptive capacity and technological acquisition.

Figure 1
Proposed model
author’s own elaboration

In this research, the dependent variable is technological innovation, which is composed of two dimensions: product innovation and process innovation (Gronum, Verreynne and Kastelle, 2012). Product innovation is the result of the sum of the dichotomous answers to the question of whether or not the firm managed to introduce to the market to: a new product, a new service, a significantly improved product or a significantly improved service. Process innovation is the result of the sum of the dichotomous answers to the question of whether or not the following were introduced: new processes or significantly improved processes.

For this study, three independent variables have been considered. Firstly, non-technological innovation is used. Following the Gronum, Verreynne and Kastelle (2012) approach, non- technological innovation has two dimensions: organizational innovation and marketing innovation. Organizational innovation is measured as the sum of the dichotomous answers to three

questions related to the activities carried out by the firm: new business practices, new methods of organizing work and new methods of organizing external relations with other firms or public institutions. Marketing innovation is measured as the sum of the dichotomous answers to four questions that are related to the following items: significant changes in the design or packaging of the good or service, new means or techniques of product promotion, new methods for product positioning in the market or sales channels and new methods of pricing goods or services.

Secondly, AC is calculated on the basis of three variables: expenditure on research, training for innovation and the R&D department. The first two variables were transformed by applying logarithm base 10, and the last one is a dichotomous variable. Thirdly, technological acquisition is made up of the following variables: machinery investment, hardware investment and software investment, all transformed by applying logarithm base 10.

The firm size and firm age can influence technological innovation. The firm size (expressed as a logarithm) is measured by the number of employees (Schoenmakers and Duysters, 2006; Laursen and Salter, 2006), and the firm age is measured by the number of years (expressed as a logarithm) from its foundation to the year in which the firm data were recorded (Thornhill, 2006).

SMARTPLS 3 software, which applies the partial structural equation estimation model in two steps, according to Chin, Marcolin and Newsted (2003), was used. First, the measurement model isestimated when determining the relationship between the indicators and the latent construct. Second, the structural model, in which the relationships between the constructs are obtained through the coefficients and the level of significance, is estimated. Hair et al. (2019) stated that SMARTPLS should be applied when the data is secondary and when the data demonstrate a lack of normality; the data for this study met both of these criteria.

The results that were obtained using descriptive statistics, the measurement model, the structural model, mediation analysis, control variables, invariance measurement and multi-group analysis are shown below.

Table 1 shows firm participation for the years 2012 and 2015, respectively, according to their size (the number of employees), their age as measured from the start of operations (before 1975, old; between 1975 and 1992, mature; and from 1992 onwards, young) and their technological intensity.

Table 1
Description of manufacturing enterprises 2012 and 2015
author’s own elaboration

Table 2 shows the outer loadings of the constructs for the years 2012 and 2015, respectively.

As can be seen, all loads are greater than 0.5, so the constructs must remain in the model.

Table 2
Outer loading on the constructs (models 2012 and 2015)
author’s own elaboration.

Table 3 shows the reliability and validity indicators for both years (2012 and 2015). It can be seen that for the Cronbach alpha (CA), the constructs have a value above 0.5. With respect to composite reliability (CR), all constructs have values greater than 0.7; the average variance extracted (AVE) is above 0.5. In addition, it can be seen that, with regard to multicollinearity, the variance inflation factor (VIF) is controlled for values of less than 5. The coefficient of determination (R2) for the relationship between absorptive capacity and non-technological

innovation is 0.556 (for 2012) and 0.409 (for 2015), and for the relationship between the following independent variables: AC, non-technological innovation and technological acquisition; and the dependent variable: technological innovation, the coefficient of determination is 0.253 (for 2012) and

0.187 (for 2015). According to Hair Jr et al. (2019), coefficient of determination values of 0.50 and 0.25 are considered moderate and weak, respectively. Based on the results of the indicators, it is possible to carry out the structural model.

Table 3
Reliability and validity indicators for 2012 and 2015
author’s own elaboration

Table 4 shows that all variables achieve discriminant validity following the criteria of Fornell and

Larcker (1981).

Table 4
Discriminant validity of 2012 and 2015
author’s own elaboration.

After evaluating the measurement models, the structural model was estimated.

Table 5 shows the coefficients and t-value for each model’s construct for the years 2012 and

2015. To generate statistical significance in the hypotheses, according to Hair Jr et al. (2014), the bootstrapping technique was used, with 4000 samples.

Table 5
Results of the structural model 2012 and 2015
author’s own elaboration

The goodness of fit index (GoF index) was used to verify the model fit (Tenenhaus et al., 2005). The GoF index ranges between the values of 0 and 1. Although there is no minimum threshold, a value greater than 0.31 is recommended (Camisón and Villar-López, 2014). For the 2012 model, the GoF index shows a value of 0.53 and, for 2015 model,

0.45. In both cases the indices are higher than the

minimum recommended to guarantee the model fit.

The analysis of the coefficients of the structural models for the years 2012 and 2015, shown in Table 5, allows for the empirical verification of the following hypotheses: for hypothesis 1 (“There is a positive and statistically significant relationship between AC and technological innovation”), it can be stated that the results coincide with those obtained in the studies carried out by Rangus and Slavec (2017) and Ali and Park (2017), which indicate that firms showing a higher level of AC at the same time show higher levels of product and process innovation capability.

With regard to hypothesis 2 (“There is a positive and statistically significant relationship between AC and non-technological innovation”), it can be specified that the results are in line with the study by Chen and Chang (2012).

Regarding hypothesis 3 (“There is a positive and statistically significant relationship between non-technological innovation and technological innovation”), it should be pointed out that, unlike the study by Camisón and Villar-López (2014), in which it was concluded that organizational innovation develops firm technological innovation

capability, this study considers non-technological innovation, which includes not only organizational but also marketing innovation.

With respect to hypothesis 4 (“There is a positive and statistically significant relationship between the acquisition of machinery, hardware and software and technological innovation”), it can be concluded that it corroborates what was pointed out by Tello (2017), who argued that Peruvian manufacturing firms prefer to innovate by buying machinery, hardware and software.

When analyzing non-technological innovation, certain steps are evaluated to confirm whether or not it is a mediating variable and, if so, whether total or partial mediation is present. According to Hair Jr et al. (2014), mediation refers to a situation in which a mediating variable in some form absorbs the effect of an exogenous construct (with independent variables) in an endogenous construct (with a dependent variable) in the PLS path model.

Table 6 shows the explained variance assessment (VAF) and determines to what extent the mediation process explains the variance of the dependent variable. The rule is that, if the VAF is less than 20 %, one must conclude that there is no mediation; a situation where the VAF is greater than 20 % and less than 80 % could be characterized as a typical partial mediation (Hair Jr et al., 2016), while a VAF above 80 % indicates complete mediation.

As noted, Table 6 shows that non-technological innovation mediates the relationship between AC and technological innovation. The analysis of the

variance indicator for the 2012 model is 40.97 %, and, for 2015, the indicator is 38 %. Therefore, in both cases, non-technological innovation partially mediates the relationship between AC and technological innovation.

This result is in line with the findings of Min,

Ling and Tan (2016), who found that organizational

innovation partially mediates the relationship between AC and technological innovation. The present model shows that not only to organizational innovation but also to marketing innovation as a component of non-technological innovation partially mediates this relationship.

Table 6
Mediation outcome for 2012 and 2015
author’s own elaboration.

Table 7 shows the coefficients, standard deviations and p-values of the control variables for the years

2012 and 2015, respectively.

Table 7
Mediation outcome for 2012 and 2015
author’s own elaboration.

From Table 7, it can be seen that the firm size has a small, negative and statistically significant coefficient. The literature points to a positive relationship between firm size and innovation (Zuniga and Crespi, 2013). However, Benavente (2006) argues that, in some cases, factors other than size, such as demand pressure, encourage firms to innovate.

In the case of the firm age, the coefficients are neither significant nor contradictory. The literature shows mixed results. Nieto, Santamaría and Fernández (2015) point out that mature firms should be more prone to innovate because of the experience they have acquired, but Cucculelli (2018) states that a negative relationship is questionable,

indicating that young firms tend to be innovative

and assume greater risks.

The invariance of the composite models should be measured before comparing the groups used in the 2012 and 2015 models. As SMARTPLS software was used, Henseler, Ringle and Sarstedt (2015) recommend using the MICOM (“measurement invariance of composite models”) procedure.

The MICOM procedure requires three steps to be carried out. The three steps are as follows: (i) configurable variance, (ii) compositional variance and equality of mean values and (iii) composite variances. Step (i) does not require statistical testing, only the verification that the data have been treated identically for both groups. Step (ii) involves

performing the permutation test. If the permutation test reveals that the correlation c (the average of the correlation obtained by permutation) is not significantly different from (i), then compositional invariance is established. In this study, 5000 permutations were carried out. Step (iii) assesses the equality of the mean values and composite variances. If the statistical test determines that

the mean values and composite variances are not significantly different, then the equality of the mean values and composite variances is established.

As shown in Table 8, in both cases, the null hypothesis is rejected, so that the averages and variances of the 2012 firms showed significant differences from those of the 2015 firms.

Table 8
MICOM model results
author’s own elaboration.

In conclusion, the results obtained, after applying the procedure for measuring invariance, conclude that the invariance is complete and, therefore, it is possible to proceed with the analysis of the two groups.

Multi-group analysis was conducted to determine the heterogeneity of the firms’ innovative behavior in 2012 and 2015. A total of 2000

permutations were used for greater robustness of the results. As shown in Table 9 and according to Chin and Dibbern (2010), two t-tests are carried out. The first t-test assumes that the variances are equal and the t-parametric (EV) indicator is obtained. The second t-test assumes that the variances are different and the t-parametric indicator (NEV) is obtained. After applying the tests, it can be seen that the results are similar.

Table 9
Multi-group comparison test results
author’s own elaboration.

As shown in Table 9, when comparing the coefficients of the models corresponding to the years 2012 and 2015, the relationships between AC and technological innovation and between non-technological innovation and technological innovation remained constant. This situation was not evident for the relationship between AC and non-technological innovation or for the relationship between the acquisition of machinery, hardware and software and technological innovation. In the latter two cases, the firms that participated in the 2012 survey made greater efforts to develop higher levels of AC and invested more resources acquiring machinery, hardware and software, and, in this way, improved their innovation capability.

This research work focuses on understanding the changes between the differences in innovative behavior of Peruvian manufacturing medium- low and low-technology firms between 2012 and 2015. Initially, the firms that participated in the 2012 innovation survey developed higher levels of absorptive capacity and increased expenditure of resources for the acquisition of machinery, hardware and software compared to those firms in the sample of 2015. Although this initial perspective could seem contradictory to the main assumptions of the model, an overall study of the results exhibits new perspectives on the evolution of innovative behavior in medium-low and low-technology firms.

The main point in this discussion starts from the evidence that the mediation effect of non- technological innovation in the relationship between absorptive capacity and technological innovation appears in the two samples. In both samples, the effect is very similar (Table 6) and shows that it is necessary to develop non- technological innovation to favor technological innovation. This work results show that this effect has not changed in two different periods. Although firms increased considerably their efforts on digital transformation this mediation effect has not reduced its importance. This behavior could suggest that the effect of non-technological innovation in technological innovation is something permanent and that opportunities in technological innovation either could come from or can be favored by non-technological innovation efforts.

Deepening the analysis of the models for each sample, two path present significant differences between the sample of 2012 and the sample of 2015. On the one side the influence of absorptive capacity on technological innovation is lower in the sample of 2015 (0.433) than in sample of 2012 (0.501). This result means that technological innovation depends less on a firms’ absorptive capacity for medium-low and low-technology firms in 2015 than for firms in 2012. Interestingly, this result can suggest that low-technological intensity firms have internalized some practices that are less dependent of firms’ absorptive capacity and proposes to explore for new factors that can influence technological innovation.

On the other hand, the path from technological acquisition to technological innovation is lower in the sample of 2015 (0.148) than in the sample of 2012 (0.312). Taking into account the increasing effort of digital transformation of all firms, the reduction on the effect of technological acquisition into technological innovation suggests a delay on the effect of new equipment and hardware on technological innovation. Also, taking into consideration the persistence of the mediating effect of non-technological innovation between absorptive capacity and technological innovation, this result suggests that acquisition of new equipment and hardware has to be accompanied of organizational and marketing changes that could mediate in the effects on technological innovation; which has been related to the commoditization of information technologies (Carr, 2003).

This research work intended to contribute to a better understanding of innovation efforts in manufacturing medium-low and low-technology firms, with focused attention on Peruvian firms. In this vein, this research work aims to contribute to this understanding by shedding some new light to the relationship of absorptive capacity, non-technological innovation and technological acquisition on technological innovation. The study’s outcomes suggest taking into consideration the persistence of the mediating effect of non- technological innovation between absorptive capacity and technological innovation, to be aware of new factors that could complement absorptive capacity, and the commoditization of the digitalization efforts of firms.

From an academic perspective, this research proposes new challenges regarding those factors or variables that can help to understand how technological or non-technological innovation can be developed in medium-low and low-technology firms. This adds to recent perspectives where organization learning has been used to understand how ERP implementation affects organizational performance in a context of digital transformation and where the impact of technology is found to have many different facets when it is adopted by small firms (Riverola and Miralles, 2016).

The development of this study makes it possible to identify some practical implications. Thus, the managers of medium-low and low-

technology firms should encourage an increase in absorptive capacity and other factors with the intention of developing more technological innovations, i.e. product or process innovation. Also medium-low and low-technology firms should allocate resources for the acquisition of machinery, hardware and software, and include those organizational changes that can accompany the implementation of new equipment to participate in developing technological innovations. Overall, decision-makers in low-technological intensity firms should consider investment efforts in new technology as an organizational change challenge and take into consideration all impacts that can affect the overall organization.

The present study is not without limitations. First is that all samples were obtained from a single source, namely the databases of the national innovation surveys of the manufacturing industry in Peru. It is suggested that future research be carried out in other Latin American economies in order to make comparisons and generalizations of the relationships that can be established between the constructs.

Second is the use of samples that include all industrial sectors with lower technological intensity. It would be very valuable to develop research in specific industries, such as the food industry, the garment industry or basic chemical products, to identify what activities firms implement to develop technological innovation in each industry.

Third is how absorptive capacity was measured. Here the criteria considered by Escribano, Fosfuri and Tribó (2009) were used and adapted to the database of the INEI of Peru. Rather, it has been suggested that questionnaires be developed to better measure the absorptive capacity construct (Fernhaber and Patel, 2012; Tortoriello, 2015).