Resource Accumulation through Economic Ties

In our new paper, Resource Accumulation through Economic Ties: Evidence from Venture Capital, which was recently published in the Journal of Financial Economics, we develop a robust and generalizable methodology that allows us to separately identify the seeking of similar versus highly or differently endowed partners. We estimate our model in a setting in which organizational networks are of primary importance: the VC industry.

Our findings suggest that concerns over agency conflicts are not a primary concern for ties among VC firms, but instead are dominated by the desire to accumulate distinct resources for the production function. Furthermore, in the VC setting, value-added resources other than capital appear to be able to exist separately from capital and still be exploited effectively. 

A key feature of our analysis is how we define resources for firms. A standard problem in the literature is that many firm characteristics observable to researchers are highly correlated and can be thought of as a proxy for some facet of unobservable or harder-to-quantify resources such as “quality.” Thus, while resources may be observed by market participants and in theory may be orthogonal, the proxies used to represent resources are often inexact and correlated. When resource trading exists and proxies for resources are positively correlated, estimating equations that do not account for resource trading will produce biased coefficients with a tendency to confirm similarity. While our methodology accounts for the possibility of resource trading, our characterization of resources provides advantages in avoiding such issues: we employ factor analysis to identify orthogonal sources of variation in firm characteristics, which we then utilize as measures of resource endowments.

In our particular laboratory of the VC industry, the underlying factors we uncover all have natural, intuitive interpretations. The characterization is parsimonious: we are able to capture 95% of the observable variation across 27 VC attributes with four linear and uncorrelated factors. Because they are constructed to be orthogonal, we can interpret them more easily as separate dimensions of variation. The first factor loads strongly on variables related to VC firm time-in-existence, both generally and as an investor in specific industry sectors, and thus appears to measure experience. The second loads strongly on various measures of network centrality, closeness and betweenness, as well as the number of companies in which a firm has recently invested, and thus appears to measure access. The third loads strongly on measures of firm assets under management, dollar volume of investment, and uninvested capital and can be interpreted as measuring available capital. The fourth loads strongly on measures of VC firm diversity of investment across industry sectors, geography, and stage of investment, and thus measures investment scope. 

In contrast to the literature on social networks among individuals, it is unclear in an organizational setting that explicit preferences for similarity would underlie the motive to tie. Similarity-based motives for organizational network ties are generally attributed to the avoidance of agency conflicts, as disparate levels of a given resource may lead to expropriation, to hold-up due to informational advantages, or to extraction of rents from different outside options.

Alternatively, firms may choose partners in an effort to aggregate particular resource endowments. They may therefore seek the highest-endowed partner, with the desirability ranking of a prospective partner independent of an organization’s own resource levels. In addition, given the existence of multiple resources, it is possible that firms also distribute resources, i.e., trade, in addition to aggregating them. Critically, for resource trading to be possible, it must be the case that agency concerns are either sufficiently small or resolvable through other means such that firms with differing endowments can engage in joint activity (e.g., one firm has more expertise and less available capital for investment, while the other firm has more available capital and less expertise). 

We find little evidence for similarity at the organizational level as the primary driver of tie formation in any of the four resources. This result is contrary to the predictions of many agency and adverse selection models, which suggest that matching should be sought based on similarity when the resource in question would allow the better-endowed partner to hold-up the less-endowed partner. Further, we find that VC firms seek partners with better network positioning and with a more generalist approach, and this preference is independent of their own positioning and approach. In contrast, economic ties appear to be enhanced by dissimilarity in experience.

Looking across resources, we find clear evidence of trading motives. Ties are more common when one partner has more available capital and the other is more experienced, has greater access, or greater investment scope. However, there is no evidence of gains from trading across value-added resources (scope, access, or experience). Thus, our model estimates suggest that net gains from trade are larger when a single firm in the pair is endowed with higher levels of all of experience, access, and investment scope and the other VC in the pair has a high endowment of available capital. Combined with our other empirical findings, it appears that capital can exist outside the firm and be combined effectively with inside value-added but the same does not apply to combinations of inside and outside value-added resources.

The full paper is available for download here.