4.2 Multivariate Data Analysis
This session presents how the explanatory variables influence the Economic Complexity Index. As mentioned, a multiple linear regression model was executed because the ECI is a metric variable.
In order to verify the quality of the regression, Fig. 2 was initially prepared, which allows comparing the predicted values of the ECI with the observed values. Visually, it is possible to verify a reasonable quality, requiring more robust analyses for confirmation.
To confirm the quality of the regression and verify the variables that influence the ECI, Table 4 is presented.
Table 4
– Multiple linear regression analysis
Variable | Coefficient | Std Error | t-statistic | P-value |
INST | -0,001972 | 0,006476 | -0,300 | 0,761 |
HCAR | 0,009963 | 0,006500 | 1,530 | 0,128 |
INFR | 0,035676* | 0,007609 | 4,690 | 0,000 |
MRKT | -0,002331 | 0,005752 | -0,410 | 0,686 |
BUSN | 0,023132* | 0,006545 | 3,530 | 0,001 |
Constant | -2,142475* | 0,271118 | -7,900 | 0,000 |
Obs. | 112 | Response variable | ECI |
Prob (F) | 0,000 | R² adjusted | 0,7708 |
* statistically significant coefficient at 1% significance |
Source: Elaborated by the authors.
The results indicate that the model is significant at 1% significance since the p-value of the F test for the model is less than 0.01, rejecting H0 that all coefficients are equal to 0. Furthermore, the model's explanatory power is 77.08%, which can be observed by the adjusted R² value, demonstrating good adequacy and good predictive power.
When analyzing the coefficients, it is verified the rejection of the null hypothesis, that the values of the coefficients are equal to 0, only for the pillars 'Infrastructure' and 'Business Sophistication', in addition to the constant term of the equation (P-value < 1%). For the other variables, the null hypothesis is not rejected; that is, it can be inferred that the coefficients are equal to 0. Therefore, the pillars 'Institutions', 'Human Capital and Research', and 'Market Sophistication' do not directly influence the Economic Complexity Index (P-value > 1%).
The coefficient of the 'Infrastructure' pillar is equal to 0.035676, which means that an increase of one unit in its score has a positive impact of 0.035676 on the nation's Economic Complexity Index. The same interpretation applies to the coefficient of the 'Business Sophistication' pillar, with a variation of one unit in its score having a positive impact of 0.023132 on the Economic Complexity Index.
These results confirm the hypotheses H3 and H5 previously presented. Confirmation of hypothesis H3, that 'Infrastructure' has a positive impact on ECI, corroborates Smith's articles (1997); Jabbouri et al. (2016), who emphasize the importance of infrastructure in general, ICTs and ecological sustainability. Confirming hypothesis H5, that Business Sophistication positively impacts ECI, corroborates the studies by Razavi et al. (2012); Audretsch and Feldman (2004), who highlight the importance of establishing partnerships, absorbing and overflowing knowledge.
Hypotheses H1, H2, and H4, previously highlighted, could not be confirmed, as it was impossible to establish a positive or negative relationship between the explanatory variables and the ECI. However, it should be noted that the development of one of the pillars inevitably impacts the others, making it impossible to achieve high scores in just one of them, for example. In this sense, caution should be exercised when interpreting these results; a qualitative analysis of how the pillars interact and how one can raise the other's score is essential.
It is also highlighted that the statistical inferences can only be made if the assumptions of the multiple linear regression model, previously highlighted, are respected. For this research, all assumptions were respected, according to the procedures presented in the methodology section, making it possible to make inferences from the results. That is, the interpretations made throughout the article are valid.
From the results of the econometric models, some discussions are worth mentioning. The 'Infrastructure' pillar, identified as the one that most influences the Complexity Index, comprises the parameters of General Infrastructure, ICTs, and Ecological Sustainability. The first two parameters have a more explicit relationship with the development of capabilities (Adamides and Karacapilidis 2020; Smith 1997), making achieving greater economic complexity possible, while sustainability deserves more attention. Couple of recent studies assess the inverse causality between economic complexity and ecological sustainability: whether economic complexity influences ecological issues (Huang et al. 2022; Shahzad et al. 2021).
The results found in these studies point, in general, to a positive relationship between economic complexity and gas emissions or environmental externalities (Shahzad et al. 2021). That is, countries with a high level of economic complexity tend to generate more environmental externalities, which could be considered controversial to the result of this article. This study found that countries with activities aimed at ecological sustainability have high economic complexity. A possible justification is that countries emitting more polluting gases must seek innovative solutions to reverse this, with policies promoting diversification and innovation (i.e., complexity) and potential solutions to ecological issues (Ikram et al. 2021). Therefore, it is suggested that, although more complex economies generate more environmental impacts, they also seek innovative solutions to this challenge.
Another important aspect is the implicit relationship between the Institutions pillar, in its three aspects (political, regulatory, and business), and the Ecological Sustainability parameter of the Infrastructure pillar. As seen previously, Infrastructure was the most relevant pillar to explain the level of economic complexity of the nation, and the Institutions pillar had no statistical relevance. However, high-quality Institutions directly influence sustainability indicators through appropriate laws, regulations, property rights, and ways to combat corruption, which can potentially reduce environmental externalities (Ikram et al. 2021; Shahzad et al. 2021). Fabrizi et al. (2024) indicate that the favorable influence of regulatory measures on both innovation and export activities exhibits an escalating trend in correspondence with a nation's pollution intensity level. This observation implies that the implementation of environmentally conscious policies, when effectively synchronized, may embody a mutually beneficial strategy, concurrently nurturing sustainability goals alongside bolstering international competitiveness. Therefore, a possible relationship between the pillars is perceived, and it is not recommended to neglect one over the other but to work together to amplify the results that more intensely influence the local economic complexity.
Although not the focus of this article, it should be noted that, besides the ecological issue, institutions can be considered the backbone between technological advancement and social inequality (Biurrun 2022), a central aspect of development. Even if the results point to a non-relevance of institutions concerning economic complexity when reflecting on social issues, it appears that this pillar should be addressed again.
In addition, based on the non-significant result for the Institutions pillar, it is believed that more than just defining norms and bureaucratic structures is needed to promote positive results for the innovation system. This result corroborates the research by Sweet and Eterovic (2019), which demonstrates, theoretically and empirically, that how ideas and innovations reach the productive sector to create more advanced systems is more important than the regulatory aspects of protection themselves.
The second pillar highlighted as relevant to the level of economic complexity of a country was Business Sophistication, directly related to the intensity and overflow of knowledge, establishment of partnerships, and innovation links. A possible justification is that knowledge spillover is directly related to increased productivity (Audretsch and Belitski 2020), an important factor for industrial complexes essential for complex economies (Dong et al. 2022; Gala 2017). Furthermore, there is evidence suggesting that companies part of local innovation ecosystems centered in innovation parks exhibit higher levels of innovation and greater technological diversification compared to their counterparts (Boyer, Ozor, and Rondé, 2021).
Also noteworthy is the geographic importance of knowledge spillover (Audretsch and Feldman 2004), which demonstrates the critical role of collective structures, such as innovation environments, industrial complexes, and local productive arrangements, among others, which intensify aspects related to the establishment of networks and partnerships, in the local complexity index. In this sense, for public policies to be efficient, intense cooperation is needed between large companies and others, especially with customers and competitors (Douglas and Radicic 2022), in order to explore the complementarities of the actors (Rong et al., 2021).
Furthermore, the overflow of knowledge can be enhanced by mapping how to create, externalize and commercialize knowledge (Audretsch and Feldman 2004). In this scenario, we have two crucial actors, teaching and research institutions and the private sector, the first for generating and transmitting knowledge and the second for bringing new solutions to society. Intensifying this interaction to generate macroeconomic results has been challenging since the first studies on innovation systems. Therefore, policies that enhance this relationship not only through regulatory and political aspects but mainly through structured processes are fundamental for the dynamism of the innovation system (Sweet and Eterovic 2019) and, consequently, for improving local economic complexity.
This critical relationship between teaching and research institutions and the productive sector leads us to a fundamental analysis of the Human Capital and Research pillar, indicated as not relevant according to the econometric model to explain economic complexity. This may be the most surprising result of the article, as the parameters of this pillar are related to the educational and research structure of the countries. A possible justification is that more than education and research is needed to change the economic level through innovation. In addition, it is believed that research needs to be directed toward problem-solving and reaching society (Mason, Rincon-Aznar, and Venturini 2020). Additionally, according to the authors, it is essential for those who pass through the educational system to have opportunities to apply their knowledge in their own country since technical skills acquired in a quality system are fundamental to converting opportunities into results of the innovation system.
In this sense, a possible relationship can be seen between the Human Capital and Research pillar and Business Sophistication, the second being relevant to determine the level of complexity of a country and the first source of 'material' for the second. Therefore, it is not recommended to abandon one pillar to the detriment of the other but to work together to enhance the results that promote greater dynamism in the innovation system and, consequently, economic complexity.
Finally, it is surprising to highlight the non-relevance of the Market Sophistication pillar, composed of factors related to access to credit, investment, competition, and scale. A possible explanation is that these factors are generalist and applied to different economic sectors, even without complexity, such as agricultural commodities and services. In addition, it should be noted that this result contrasts with empirical evidence that indicates that the results of innovations in large companies have a more positive impact than in others (Piekkola and Rahko 2020).
From hole analysis, it is clear that one should not neglect any of the pillars but work them together. It is possible to verify those that directly influence economic complexity, being a good source of information for strategies and public policies to be defined in pursuit of development focusing on economic complexity. However, the dynamics of innovation is a complex phenomenon, with several dependent processes that require the interaction of numerous heterogeneous agents (Antonelli 2009), making it difficult to separate the pillars and roles of the actors.