The Impact of Venture Capital Monitoring

It is often argued that venture capital (VC) plays an important role in promoting innovation and growth. Consistent with this belief, governments around the world have pursued a number of policies aimed at fostering local venture capital activity. The goal of these policies has been to replicate the success of regions like Silicon Valley in the United States. However, there remains scarce evidence that the activities of venture capitalists actually play a causal role in stimulating the creation of innovative and successful companies. Indeed, venture capitalists may simply select companies that are poised to innovate and succeed, even absent their involvement. In this case, efforts by policy-makers to foster local venture capital activity would be misguided. In our paper, The Impact of Venture Capital Monitoring: Evidence from a Natural Experiment, which was recently made publicly available on SSRN, we examine whether the activities of venture capitalists do indeed affect portfolio company outcomes.

Identifying the effect of venture capital is difficult for several reasons. First, in many cases, data are only available for VC-backed companies; thus there is no control group available to estimate a counterfactual. Second, even when data on seemingly comparable non-VC-backed companies are available, it is likely that these companies differ along unobservable dimensions that may drive differences in outcomes. This is especially true given that only coarse information can generally be observed about these privately held companies. Finally, expectations about future events may drive investment, leading to reverse causality concerns (Rin et al., 2011). For example, VC-backed companies may grow faster than their matched peers subsequent to investment; however, this may simply reflect the fact that VCs seek investment opportunities with the potential for near-term rapid growth. While the ideal experiment to identify the effect of venture capital would likely involve the random allocation of VC funding, finding a natural experiment that approximates this ideal has been difficult.

We attempt to make progress on this important question by taking a somewhat different approach. Rather than seeking exogenous variation in venture capital investment, we instead exploit a natural experiment that leads to exogenous variation in monitoring costs within existing VC-company relationships. If differences in outcomes for venture-backed companies are driven only by selection, reductions in monitoring costs subsequent to investment should have no effect. On the other hand, if VC activities do matter, reductions in the cost of monitoring should translate into better portfolio company performance by allowing VCs to engage in more of these activities. For example VCs may be able to spend more time advising and shaping senior management, providing access to key resources, and aiding in company professionalization in myriad other ways that have been documented in the literature.

The shock to monitoring costs that we utilize is the introduction of new airline routes that reduce the travel time between venture capital firms and their portfolio companies. Previous work suggests that such travel time reductions lower monitoring costs for firms with headquarters that are geographically separated from their production facilities (Giroud, 2013). In the context of venture capital, there is ample anecdotal evidence that venture capitalists are sensitive to distance and travel time. For example, in response to a new United Airlines flight between Raleigh-Durham and San Francisco in 2012, the president of the Durham Chamber of Commerce stated that the new route would be valuable to “venture capitalists who like to be a direct flight away from any company they’re going to invest in” (News & Observer, August 12, 2012). Similarly, the lack of direct flights to Indianapolis is seen as an impediment to venture capital in the area: “Layovers and complicated connections are aggravating […] That’s an important consideration because most venture capitalists want to keep close tabs on the companies they invest in, which requires frequent in-person visits”(Indianapolis Star, October 8, 2000).

Consistent with the anecdotal evidence, the academic literature shows that VC activity is sensitive to distance and travel time. For example, Lerner (1995) finds that VCs are more likely to sit on boards of geographically proximate companies. Chen et al. (2010) find that VCs are more likely to invest in a distant region if they already visit one portfolio company in the same region, arguing that the time associated with monitoring a distant investment affects the decision to invest. Bengtsson and Ravid (2009) find that VC contracts are more high-powered as geographic distance increases, indicating that monitoring costs increase with distance. The inclination to invest locally is not surprising given that, according to survey evidence, venture capitalists spend most of their time managing portfolio companies, and frequently visit company sites (Gorman and Sahlman, 1989; Sahlman, 1990).

We begin by documenting that there is indeed significant venture capital activity outside of the three main metropolitan areas of San Francisco, Boston and New York. Indeed, approximately 50% of both venture-backed companies and venture capital investment firms are located outside of these three regions. This is consistent with the findings of Chen et al. (2010). Moreover, we show that it is not uncommon for VCs to invest in distant portfolio companies. Given these patterns, we then explore how the introduction of airline routes affects aggregate venture capital flows between MSAs (Metropolitan Statistical Areas) in the United States. Using a difference-in-differences estimation framework, we find that the introduction of a new direct airline route leads to a 5.3% increase in total venture capital investment, and a 3% increase in likelihood of VC activity. This regional analysis also allows us to explore whether the overall increase in investment is driven by increases on the extensive margin, intensive margin, or both. We find that both first time investments as well as follow-up investments increase with reductions in travel time. In the case of new investments, this could be driven by reduced screening costs in addition to the aforementioned monitoring costs.

A natural concern is that local shocks, either in the source or target MSA could be driving the results. For example, a booming local economy may lead to the introduction of airline routes and also increased VC investment. In this case, we may estimate a spurious positive effect of travel time reductions on investment. However, since our treatment is defined at the MSA-pair level, we can control for such local shocks. Specifically, we include MSA by year fixed effects for both source and target regions in all the regressions.

New investments that occur after the treatment may tend to have different outcomes, not because the reduction in travel time affects the level of involvement of the VC, but because it changes the selection process. For example, VCs may have a lower hurdle rate for investments in a distant region subsequent to the introduction of a direct flight to that region. Therefore, in order to isolate the effect of venture capital involvement, the remainder of the analysis focuses on VC-company relationships that existed before the treatment. This also means that we move from doing analysis at the MSA-pair level (henceforth, “regional analysis”) to analysis at the VC-company pair level (henceforth, “relationship analysis”). For the relationship analysis we again use a difference-in-differences estimation framework, controlling for local shocks in both the VC and company regions.

The primary outcomes we examine are the quantity and quality of innovation (as measured by the patent count and citations per patent, respectively), as well as ultimate success (as measured by exit via IPO or acquisition). We find that the introduction of a new airline route that reduces the travel time from a lead VC to a portfolio company leads to a 3.74% increase of in the number of patents the portfolio company produces and a 3.54% increase in the number of citations per patent it receives. Further, the treatment increases the probability of ultimately going public by 1.23%, and of having a successful exit (via IPO or acquisition) by 2.79%.

Next, we investigate the channel through which these effects operate. Our main hypothesis is that a reduction in monitoring costs should increase VC involvement, which may in turn improve portfolio company performance. Unfortunately, we cannot directly observe whether VC involvement actually increases when monitoring costs decline. However, we take advantage of the fact that involvement for certain VCs should be more sensitive to changes in monitoring costs than for others. Specifically, VCs often syndicate their investments, and when this occurs, one VC typically takes the role of the lead investor. The lead investor is generally more actively involved in the monitoring of the portfolio company, while others act more as passive providers of capital. Indeed, Gorman and Sahlman (1989) find that venture capitalists acting as lead investors spend significantly more time on their portfolio companies than they would otherwise. Given that lead VCs play a greater role in monitoring, their monitoring effort should be more sensitive to reductions in monitoring costs as should portfolio company performance. Consistent with this argument, we find that our results are driven primarily by reductions in travel time for lead VCs rather than other members of the investment syndicate.

We further verify the robustness of our results in several ways. First, it is possible that if a portfolio company is performing very well, a new airline route may be introduced in response. While we do not believe this to be likely, it would bias our estimates. To ensure that such pre-existing trends are not driving our results, we examine the dynamics of how company outcomes change in the years surrounding the treatment. We find the bulk of the effect coming 1 to 2 years after the treatment, with no “effect” prior to the treatment. Second, reductions in monitoring costs should be greater the greater the reduction in travel time. Consistent with this argument, we find larger effects associated with larger travel time reductions. Third, we show that the results are robust to considering only new airline routes that are the outcome of a merger between two airlines or the opening of a new hub. Such cases are likely to be even more exogenous to any given VC-company pair. Finally, we show that the results are robust to a host of alternative constructions of the outcome variables.

The full paper is available for download here.