Learning by Investing: Entrepreneurial Spillovers from Venture Capital

The academic literature on entrepreneurial finance has largely treated investors and entrepreneurs as distinct identities. But the boundaries between investors and entrepreneurs have become increasingly blurred, primarily due to the growing importance of angel groups, crowdfunding, venture funds, and other types of financial intermediation.

In the setting of venture capital (VC), we have also witnessed a similar trend with abundant anecdotes. For instance, in 2009 Jitendra Gupta, an entrepreneur based in San Francisco, founded Punchh, a loyalty platform for restaurants, groceries, retailers, and convenience stores. After seven years as CEO, he stepped aside (Punchh was acquired for 500 million dollars in 2021). Among other activities, he invested in the KAE Capital Fund II in 2016, a fund that mainly focuses on investments in the consumer products, healthcare, and software sector. Two years later, he started a new venture, MyYogaTeacher, an online and interactive platform connecting students with yoga experts.

Though numerous studies have highlighted how VC adds value to portfolio firms (e.g., Kortum and Lerner, 2000; Hellmann and Puri, 2002; Bernstein et al., 2016), it has received much less scrutiny on how the experience as investors, i.e., limited partners (LPs), and entrepreneurs interact with each other. This paucity of research is striking given the anecdotal evidence of the importance of this channel for VC influence. As highlighted on AngelList’s website, “Aside from returns, benefits to becoming an LP [investing in VC funds] might include… access to information … [and an] expanded network.” Both of these are key resources for entrepreneurs.

In our paper, “Learning by Doing: Entrepreneurial Spillovers from Venture Capital,” we study how being an LP in a VC fund affects an individual investor’s subsequent entrepreneurial activity. While this is certainly not the only setting where we might expect knowledge transmission from investing, it is one where there are knowledgeable intermediaries that might facilitate learning, as opposed to settings such as crowdfunding.

We assemble a unique dataset that covers all firm creation and domestic VC activity in China. It combines the proprietary administrative business registry data (from the State Administration for Industry and Commerce, or SAIC) with the VC fundraising and investment records from Zero2IPO and the Asset Management Association of China (AMAC). Our data contain the entirety of firm creation activities, the shareholders of these firms, VC equity investments, and the names and financial commitments of all limited partners from 1999 to 2018. This dataset’s availability can potentially expand the research agenda about limited partners in VC funds.

Our empirical investigation starts by comparing the entrepreneurial outcomes of individual LPs in VC funds that successfully launched to those of individual LPs in funds that failed to launch. We find that after investing in a successfully launched VC fund, individual LPs create significantly more ventures than do LPs in funds which failed to launch. To address the concern that fund’s failure to launch is not random, we use the fraction of total committed capital from corporate LPs that encounter industry distress in months before the fund’s approval at the SAIC as an instrument for the fund’s failure to launch. This delivers a robust result: After becoming an LP of a successfully launched VC fund, the average number of new ventures started by the individual LP per year will increase by about 0.07, which is about 17% of a standard deviation. Our estimate implies that on average an individual LP will create one more startup in 14 years after investing in VC, relative to their counterparts in the failed-to-launch funds. The result also holds if we employ as a dependent variable the total number of ventures launched after becoming an LP.

What do these newly created ventures look like? We compare the characteristics of ventures created after investing in a VC fund (new ventures) to those created prior to investing in VC (old ventures). We find that the new ventures are especially more likely to be in high-tech service industries, an area more popular with venture investors. We find that new ventures on average file more patents within three years after being founded compared to the old ones within the same period among individual LPs in the successfully launched funds relative to those in the failed-to-launch funds. However, we do not find that they hire more employees, suggesting a higher innovation efficiency within the new ventures.

In a final set of results, we explore the underlying channels explaining the entrepreneurial spillovers to individual LPs. A learning channel might contribute because individual LPs access superior information about VCs’ portfolio companies. Alternatively, a financial constraints channel, through which investments in VCs can result in a substantial financial return to LPs and help ease the financial constraints facing entrepreneurs, could explain the pattern. The network channel, as one might expect that interacting with GPs makes it easier to obtain VC financing for LPs’ own ventures and incentivizing more firm creation, could also explain the pattern. We find supporting evidence for the learning channel, but not for the financial constraints channel or the network channel. Specifically, we find that the industry and patent classification codes of new ventures share more similarity, relative to the old ventures, to those of VCs’ portfolio companies. The entrepreneurial spillover effect after investing in VC is strongest for LPs investing in funds managed by “better” quality GPs or for first-time LPs, again consistent with learning. Inconsistent with the financial constraints channel, we do not find a significant increase in individual LPs’ entrepreneurship after investing in VC funds with a successful exit in their portfolios. Contrary to the prediction of the network channel, we do not find a significant increase in VC financing of the new ventures created by LPs.

Our findings suggest a positive entrepreneurial spillover from investing in VC funds, implying a learning-by-investing mechanism. Yet we conclude with one caveat. The Chinese and U.S. VC markets are not precisely comparable. But LPs’ involvement with funds anecdotally appears similar in many aspects across both countries. GPs often engage their LPs quite actively, such by requesting advice and introductions. Regardless of fund location, LPs often go above what is contractually required of them in their partnership agreements. Therefore, the learning-by-investing mechanism documented in our paper extends well beyond the borders of China into more developed VC markets.