![Sample our Global Development journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5929/TOC-Subject-Sample-2021-Global-Development.jpg"})
Abstract
This study extends theory on the effects of public policies stimulating technology demand and of industry network ties on firm-level entrepreneurial behaviour. We take an institutional perspective to develop a theoretical model examining the mechanisms through which public policies, regulatory uncertainty, and industry network ties affect firm-level entrepreneurial decision-making processes and the ability to introduce highly innovative products and to sustain superior performance. We focus on firm-level effects, which enables the study of the tension between institutional pressures of homogeneity and competitive pressures of heterogeneity for entrepreneurial decision-making processes in environments characterized by policy-induced market demands. To test our hypotheses, we draw on data from a large-scale survey among German renewable energy firms. Our results show that public policies can constrain firm innovativeness and risk-taking behaviour because they steer firms towards a more conservative attitude and discourage the pursuit of high-risk innovation projects. However, firms can counteract these influences and enhance their innovativeness by maintaining close network ties with research associations as we find that innovativeness and a highly innovative product portfolio are key success factors. In summary, these findings provide important implications for the study of public policy effects, industry network ties and entrepreneurial behaviour.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
* An earlier version of this paper has been presented at the 73rd Academy of Management Annual Meeting.
1. We focus on individual firm responses to account for the fact that decision-makers react more strongly to their perception of environmental factors rather than to its objective features (Lewin Citation1951). Hence, even if firm managers are subject to similar institutional forces, they might perceive and respond differently (Delmas and Toffel Citation2004).
2. We apply institutional theory as sociological institutionalism or new institutionalism, which was developed to explain organizational behaviour (DiMaggio and Powell Citation1991; Meyer and Rowan Citation1977).
3. The concept of an organizational field is similar to that of an industry, but more inclusive. Whereas the organizational field includes all relevant firms and organizations, an industry typically refers to a group of firms that ‘use similar inputs and technologies, produce similar products, and serve similar customers’ (Low and Abrahamson Citation1997, 440).
4. In 2000, hydro power was the most dominant source (4.3%), followed by wind (1.6%). In 2014, the percentages for individual technologies were as follows: wind: 9.2%, biomass: 6.8%, hydro power: 3.1%, photovoltaic: 5.6%.
5. In the solar thermal industry, the most important policy measure similar to the feed-in-tarriff (EEG) is the ‘Marktanreizprogramm’ (MAP), which has been part of the ‘Erneuerbare Energien Wärmegesetz (EEWärmeG)’ since 2009.
6. The most important trade association is the German Renewable Energy Federation, as well as different associations for the different sectors (e.g. German Wind Energy Association for wind energy, German Solar Energy Association for photovoltaic and solar thermal).
7. For diversified firms that operate in more than one renewable sector or other unrelated industries, we asked the CEOs to complete the survey and provide information for their most important renewable business unit in terms of sales volume.
8. Fifty-five of the one thousand two hundred and eight firms responded by saying that they were no longer active in the renewable energy sector or solely focused on activities outside of Germany. Thus, we effectively reached 1153 firms.
9. These firms were excluded from the total sample of 1208 German renewable energy firms.
10. As we are interested in analysing the impact of regulatory uncertainty, we had to reverse the data prior to the analysis.
11. The value for the average variance extracted is slightly below the threshold of 0.5. However, as the other indicators point to good and adequate reliability and validity, we decided not to further modify the construct as it is a standard measure that has been applied by influential research and because the thresholds are not equatable with fixed cut-off values.
12. We applied the Spearman Correlation (rs = 0.25, p-value < 0.01) because our sales growth variable strongly deviates from the assumption of normal distribution, as indicated by a Kolmogorow–Smirnow test and values for skewness of 3.14 and kurtosis of 13.76.
13. We applied two-sided t-tests as well as a Welch test. Despite the small sample size of 25, our variables do not deviate from the assumption of normal distribution, as indicated by a Kolmogorow–Smirnow test and values for skewness <2 and kurtosis <7.
14. More information on the results of the exploratory and confirmatory factor analysis can be provided by the authors upon request.
15. The international orientation was measured by asking the firms about the percentage of sales (2009–2011) their renewable business unit realized outside of Germany through exports, subsidiaries, cooperation, alliances, etc.
16. We applied the renowned six-item construct provided by Jaworski and Kohli (Citation1993) to measure competitive intensity. After the exploratory factor analysis, we reduced the measure for competitive intensity to four items. The results pointed to good reliability and validity (α = 0.80, factor loadings ≥ 0.54, FR = 0.79, A.V.E. = 0.50).
17. In the repeated indicators approach, which is also known as the hierarchical components model, the second-order factors are measured by the observed variables for the first-order factors. This approach is useful for formative first-order, formative second-order constructs when the number of indicators for the first-order factors is equal and if there are only minor threats to indicator collinearity (Albers and Götz Citation2006; Chin, Marcolin, and Newsted Citation2003). As stated above, we did not find indications of a threat of indicator collinearity for the first-order constructs technology and market newness that are both measured with four items.
18. We assessed the potential threat of multicollinearity between the influence of public policies and regulatory uncertainty. However, we found no major threats when looking at the correlation (rbp = 0.14, p-value = 0.10) and the variance inflation factor (VIF = 1).
19. The values were the following: proactiveness: 0.19, risk taking: 0.03, innovativeness: 0.14, degree of innovation: 0.14 and financial performance: 0.07 (see also Figure ).
20. We applied t-tests and the non-parametric Mann–Whitney-U test to assess significant differences for the control variables that were split into two groups (diversification, international orientation, competitive intensity, access to capital, operation in more than one renewable energy sector and R&D intensity). Where we had more than two groups (age, size, geographic location and the type of renewable sector), we used the Scheffé test or the Games-Howell test in the case of heterogeneity of variances, and the non-parametric Kruskal Wallis test to assess significant differences between groups.
21. When describing our results in the previous section, we highlighted that the effects on risk taking should be interpreted more carefully because of the R² of 0.08 in the basic research model. However, when integrating the moderating effect of high R&D intensity, the variance explained increases to an adequate level that allows interpretation (R² = 0.33).
22. Our final measure consisted of demand-pull policies that stimulate technology demand, as those measures promoting technology development (e.g. R&D grants) had to be dropped after the exploratory factor analysis. When integrating the latter measurement as a separate construct into our research model, we found no significant effects on the dependent variables.