A neo-configurational institutional analysis of international venture capital attractiveness and performance: insights for Asia-Pacific

Abstract

Prior research suggests that market-supporting institutions are attractive to international venture capital (IVC), yet paradoxically VC firms increasingly invest in countries characterized by institutional weaknesses. In this study, we explore the institutional contexts that make countries attractive to IVC investment and favorable to IVC performance, thus illuminating the implied risk-reward tradeoffs in IVC firm location choice and ability to navigate institutional weaknesses. Utilizing fuzzy-set Qualitative Comparative Analysis (fsQCA), we find four distinct configurations of institutions associated with IVC attractiveness. Each configuration allows for some institutional weaknesses, enabling us to elaborate theory on institutional complementarities and substitutions that drive IVC investment. Further, we find that IVC firms have performed well even in countries with pervasive institutional weaknesses, suggesting that IVC firms may uniquely contribute to the success of new ventures in such institutional contexts. We discuss implications for theory and offer policy and managerial recommendations particularly relevant to IVC activity in the Asia-Pacific region.

Appendix 1: Robustness tests

We first conducted an analysis by combining rather than stacking the data. That is, we took an average of the early and late institutional measures and summed the two outcome variables. We then recalibrated this data. The new analysis for IVC investment yielded three configurations that are in subset relationship with the four in the main analysis. That is, each of the four configurations from the main analysis is a subset of at least one of the three configurations in this robustness test. The converse is also true. Coverage of these three configurations is 0.67 and consistency is 0.86. For the IVC performance outcome, the new analysis was somewhat different and had far better coverage. Five configurations were identified that covered 0.64 of the high performing countries and had consistency of 0.87. Each of the three configurations from the main analysis is a subset of exactly one of the three configurations in this robustness test. Of course, the converse is not true, as two configurations that were not in subset relationship with the main analysis for performance were identified.

We also ran a second set of robustness analyses in which we divided the sample into the two time periods (pre- and post- financial crisis of 2008) and analyzed each time period separately to examine whether there is any significant change as a result of the financial crisis. For the IVC investment outcome, the early time period contained three sufficient configurations and the later time period contained four configurations. C1A and C2A were exactly duplicated in both periods. C3a was in subset relationship with a configuration identified in the first period and was exactly duplicated in the second period. C4a was in strong subset relationship with a configuration in the later time period but not the earlier period. In both periods, coverage and consistency scores were slightly higher than in the main analysis. For the IVC performance analysis, both time periods identified three sufficient configurations, and all three configurations from the main analysis (C1p, C2p, and C3p) were in subset relation with single (different) configurations in each of the time periods. Again, coverage and consistency scores for the performance outcome were similar to the primary analysis. Overall, there were marked similarities in both time periods for each outcome.

As a robustness analysis for attractiveness, we also used an alternative outcome. Rather than IVC investments per capita, we used IVC investments per real GDP. The results were similar to the main analysis. C1a and C2a, the two largest configurations, were exactly supplicated. Additionally, this analysis identified a third configuration containing the presence of four important institutions: a robust public governance, openness to competition, effective labor markets, and local venture capital availability. The coverage (0.62) and consistency (0.78) were moderately reduced form the main analysis. These results should probably have the greatest confidence in Configurations 1a and 2a.

We also conducted robustness analyses for each outcome consistent with the recommendations of Schneider and Wagemann (2012). Namely, we examined whether using stricter consistency and frequency thresholds leads to the detection of any configurations “that are not in a subset relation” with the main analyses (Schneider and Wagemann 2012: 286). To test this possibility, and similar to García-Castro et al. (2013), we repeated the analysis using (1) a consistency threshold of 0.85 instead of 0.80 and (2) a frequency cutoff of 2 rather than 1.

For IVC investment, the 0.85 consistency threshold, leads to two configurations with strong analogs that are in subset relation to C1a and C2a. The smaller configurations, C3a and C4a do not find analogs. The coverage of this analysis decreases to 0.59 and the consistency increases, incrementally, to 0.89. When the frequency cutoff is increased to 2, the main analysis is exactly duplicated. Each of the first three configurations (C1a, C2a, C3a) from the main analysis has at least one corresponding configuration with which it is in subset relation. The smallest configuration, C4a did not find an analog. Solution consistency and coverage were highly similar (± 0.01 in both cases) to the main analyses. Overall, this indicates that the main results, with regard to VC investment, are robust.

For the investment performance outcome, when the consistency threshold was set to 0.85, two configurations emerged. These configurations each had one analog in the main analyses (C1p and C2p), to which they were in strong subset relationship respectively. The solution coverage of that analysis dips slightly from 0.29 to 0.27 and the consistency improves slightly from 0.82 to 0.84. With a frequency cutoff of two, the main analysis was exactly duplicated again. Overall, the analysis for IVC performance were robust were and much in line with Schneider and Wagemann’s (2012) recommendations.

Finally, we conducted a series of regression analyses with the causal conditions from the fsQCA model for attractiveness as well as the (1) GDP growth rate and (2) per capita income as controls. Only public governance robustness has a significant positive effect prior to adding interactions. We then used the fsQCA solution to ascertain which causal conditions are most likely to interact with one another. The three variables that are core conditions that hang together in the main analysis for attractiveness are public governance quality, openness to competition, and product market efficiency. So, we attempted two-way and three-way interactions with these variables. However, the interactions were not significant with the exception of a two-way interaction between openness to competition and product market efficiency. Overall, these results do not provide the insights into the complementarities within and substitutions across configurations and the alternative pathways across configurations that our model uncovers, thus highlighting the value of our configurational approach.