I.3. El comercio itinerante y el comercio adyacente a la Plaza- Plaza-Mercado
I.3.2. Sobre la ordenación del comercio itinerante y el comercio adyacente a la plaza mercado
The following Table 7 illustrates the estimation results of the fixed effects treatment for a three-level set-up. Before embarking on fully fledged multilevel models, it is worth
106 For more information on the analysis of the data distribution and the completion of outlier analyses and data transformations, see section 6.6 of this present chapter.
examining the data by carefully inspecting the correlation matrix for all variables at each level of analysis. Due to multicollinearity issues, the variable rule of law is removed at regional and at national level. Based on studying measures such as variance inflation factors, additional multicollinearity issues are detected which can be ascribed to some Level3 covariates, namely commercial entrepreneurship: national, informal capital: national and social trust: national. As a result, these national-level covariates are dropped. An extensive description on the robustness checks and the rationale behind dropping certain Level3 covariates can be found in section 6.6 below.
Table 7 Fixed effects estimations.
population density: established entrepreneurial culture at regional-level (Hypothesis 1) exerts a positively significant impact on social enterprises’ employment growth (Model 1b: 0.07*)107 but no significant association is observed between commercial entrepreneurship rates and an increase in either social impact rates (Model 3b: 0.07) or revenue growth (Model 2b: 0.17).
Hence, Hypothesis 1 is only partly confirmed. Furthermore, it was tested whether access to financial resources (informal capital) positively influences social entrepreneurship growth (Hypothesis 2). In line with the prediction of this study, sufficient funding has a positive impact on employment growth (Model 1b: 0.13**), revenue growth (Model 2b: 0.14**) as well as on social impact development (Model 3b: 0.06**). For an increase of the regional funding supply by 1%, the estimated employment growth rates increase on average by 13.88%, in the case of revenue growth by 15.03% and it positively affects social impact development by 0.06 units. Next, the effect of social capital, using the instrumental variable social trust, is examined (Hypothesis 3). According to the results, social trust is positively related to social enterprises’ employment growth (Model 1b: 0.03*), revenue growth (Model 2b: 0.02**) as well as social impact development (Model 3b: 0.05**), implying that mutual trust within society is a strong driver of social entrepreneurship development at regional
107 The estimation results have to be exponentiated as the dependent factors are log-transformed variables.
Hence, if the regional commercial entrepreneurship rate increases by 1%, the expected employment growth rate is estimated to increase on average by 7.25%.
level. The supply-side Hypothesis 3 can thus be confirmed. Similarly, the size of the non-profit sector (Hypothesis 4) positively affects social enterprises’ revenue growth (Model 2b:
0.04**) as well as their social impact development (Model 3b: 0.09**). In the employment growth model, a positive but insignificant effect is noted (Model 1b: 0.03). Hence, Hypothesis 4 needs to be partially confirmed, namely in Model 2b and 3b.
With regard to adverse societal conditions at regional level (Hypothesis 5), poverty rates affect social enterprises’ development in terms of an increasing number of employees (Model 1b: 0.03*) and with respect to social impact development (Model 3b: 0.04*). No impact association can be found between high poverty rates and revenue growth (Model 2b: 0.01).
Also, the impact of national poverty rates is only positively significant in the employment growth specification (Model 1b: 0.01*). In line with Hypothesis 6, public health expenditure at regional and at national level is negatively associated with social enterprises’ employment growth (Model 1b: -0.13**; -0.12*), revenue growth (Model 2b: -0.18*; -0.15*) and social impact development (Model 3b: -0.08*; -0.04). Hence, Hypothesis 5 can be partially confirmed when testing it at regional level; Hypothesis 6 can be confirmed at both higher levels of analysis.
Two control variables are introduced at regional and at country level: Per capita GDP and population density. The estimates suggest that wealthy regions create significant demand for social entrepreneurial services and products, thus, inducing an increase in employment and revenue growth rates (Model 1b: 0.02**; 0.02* and Model 2b: 0.22**; 0.06*). However, GDP per capita has no effect on enterprises’ creation of social impact. Population density is solely positively related to revenue growth at regional level (Model 2b: 0.03**), but shows no other effect on social enterprises’ growth.
Turning to firm-level predictors, it can be observed that social enterprises’ choice upon the geographical scope of operation (Hypothesis 8b) has no effect on their growth whatsoever.
Social networks, on the other hand, are crucial for a social enterprise’s dynamism (Hypothesis 8c): Informal social networks have a positive effect on social enterprises’
employment (Model 1b: 0.04*) and revenue growth rates (Model 2b: 0.05*) as well as on their social impact development (Model 3b: 0.06**). Formal networks, however, only exert a positive and significant effect on social impact development (Model 3b: 0.05*). Furthermore, testing for operational strategies (Hypothesis 8a) leads to mixed results. Social enterprises that implement more than one business model (diversification strategy) do not experience
higher growth rates. However, the combination of several and different business models (complexity strategy) is expected to lead to higher employment (Model 1b: 0.09*) and revenue growth rates (Model 2b: 0.16*) as well as greater social impact development (Model 3b: 0.07*). Beyond the specific hypothesis at firm level, it is notable that some operational business models exert an influence on social enterprise growth. For example, social enterprises that adopt either the employment or the cooperative model (opmo2), experience on average 9.42%108 higher employment growth rates (Model 1b: 0.09*) and they achieve on average a 17.35% higher revenue development (Model 2b: 0.16**) compared to social enterprises opting for a different model.
Control variables at firm level suggest a negative impact of enterprises’ age on employment growth (Model 1b: -0.01*) and revenue growth (Model 2b: -0.07*), but a positive effect on social impact development (Model 3b: 0.13*). Thus, enterprise maturity and experience is a pivotal variable to scale social impact. Moreover, the amount of assets in 2008 is positively related to all types of social enterprise growth (employment, revenue and social impact) and social enterprises operating in the service sector (dummy variable: nace) are more likely to achieve higher revenue growth rates.