Almost a year ago, a thorough literature review was released, A Survey of Venture Capital Research, that looked at two decades of academic research on all dimensions of venture capital. This included outcomes, returns, the structure of VC firms, and so on. It is a very good review and, in light of several items that have been published over the past year, below is my own reduction of the review. This is not a comprehensive summary of the paper, nor does it include every single point or study conclusion. It is simply my own extraction of what I found to be the most useful and interesting findings from VC research. Some of their points also pertain to general research on entrepreneurship.
Interesting and Useful Findings about Venture Capital
In 1980, the amount of VC investment in US was $610 million; by 1990, it was $2.3 billion; by 2010 it was $30 billion. On average, over 200 new VC funds have been created each year since 2000. Europe and Asia now account for roughly half of total investment flows.
There are data availability and access issues, including with the notion of 'failure,' because many "bankrupt companies are retained as 'living dead,' i.e. they appear to be active rather than shown to have failed." (Because this is an issue not just with VC data, but also with Census data, this is likely a problem for other entrepreneurship data we use.) Exits are also "poorly measured" in most databases, and some researchers supplement Census LBD data with acquisition data from other sources--the inability to see M&A in Census data is another issue for entrepreneurship research.
"We would argue that in the area of VC, all databases have some sample selection issues. This is because it is conceptually impossible to clearly define a company's birth date. The most common measure is the date of incorporation, but there are several problems with this. Entrepreneurs typically engage in substantial economic activities prior to incorporation, such as doing research, developing a business plan, or raising funds. ... In some cases the incorporation is also based on some unobservable prior entity, such as a sole proprietorship. Moreover, if one takes the whole sample of incorporated companies, these include a large number of entities, such as tax shelters, that cannot be considered proper companies. ... The Kauffman Firm Survey, for example, uses the data of incorporation. Commercial databases typically capture companies at the time of the first VC round." (emphasis added)
Many VC research papers seek "to identify some causal effect of VC financing. ... A central concern here is the distinction between selection and treatment effects. VC transactions involve at least two parties, so that selection effects may pertain to both the VC and the entrepreneur sides of the market. ... there may be unobserved heterogeneity in underlying attributes of both parties that might affect the matching of entrepreneurs with VCs." There are also issues with reverse causality, "where expectatios about future events may drive agent's actions, like the expectation of a quick IPO may convince the VC to invest rather than investment leading to a quick IPO."
An important source of deal flow is spin-offs, what some researchers call "entrepreneurial spawning" from big public companies. Here, state policy on things like the enforcement of non-compete agreements is very important.
Studies estimate that during the 1990s roughly 10% of VC-backed founders were serial entrepreneurs, with serial entrepreneurs more likely both to obtain VC as well as better valuations. Prior success, not just experience in founding, appears to matter for subsequent entrepreneurial success.
"Overall, we would argue that while there is some recent research on spin-offs and serial entrepreneurs, we still know relatively little about the creation of ventures that demand VC finance. Somewhat surprisingly, there seems to be no dialogue between the VC literature and the labor economics literature on the decision to become an entrepreneur."
From 1981 to 2005, an estimated 0.11% of new companies were funded by VC; this rose to 0.22% from 1996 to 2000. In the Kauffman Firm Survey, the percentage is also less than one percent. VC investments are concentrated in a handful of high-technology sectors with high-growth opportunities--some argue that "while VCs naturally focus on industries where information asymmetries are high, within these industries they prefer to finance firms that have relatively milder information challenges."
Nevertheless, research finds that VC-backed companies comprise anywhere from 5 to 7 percent of employment in the US, up from 2.7% in the early 1980s. For IPOs, one-third were VC-backed bewteen 1980 and 2010.
VC provide equity finance while banks use debt finance, and VCs provide value-adding services. In one model, banks screen imperfectly and thus rely on collateral in exchange for capital. Entrepreneurs will thus seek VC when they have little collateral and when they need larger amounts. There is, however, "surprisingly little empirical evidence on the choice between VC and bank funding."
There is disagreement about the relative value-adding services of angels and VCs. One study finds a "strong positive correlation of angel financing on the growth and survival of companies." Other researchers find that angel and VC deals enjoy similar success rates when the investment amounts are smaller, but that "angel-backed companies are more likely to become 'living dead.' Pure VC deals have a higher success rate when they involve larger sums. ... Overall we would say that the trade-offs between VC, banks and angel financing still remain poorly understood."
Some research finds that "generalized trust explains deal formation" after controlling for investor and company effects. Social capital appears to matter a good deal, but it is unclear "how much these social criteria substitute or complement the characteristics of the underlying business opportunity." One study, comparing the relative importance of the idea and team, argues that, for VC financing at least, "ideas matter more because companies' strategies change rarely, whereas management turnover is common."
Studies have examined the use of convertible preferred equity (CPE), where investors have an option to convern debt to equity. "Investors benefit from the preferred terms when the exist value is low (i.e., on the downside) but convert to common equity when the exit value is high (i.e., on the upside). ... The optimal CPE is such that the VC only converts if the entrepreneur has provided the optimal level of effort." Others justify CPE based on taxation.
Three sources of risk for VCs are identified: internal risk, concerning the abilities of founders and the difficulty of monitoring their actions; external risk, concerning the business environment, and execution risk, concerning the venture's reliance on the founders and any potential for hold-up. The risk mixture determines the level of control rights and equity stakes.
VCs sit on boards of directors, mentor founders, help raises additional money, recruit management, and offer strategic advice. Geographic proximity affects not only investment terms but also the level of involvement. VC has been found to lead to a "pattern of professionalization" within the companies in terms of hiring practices and stock option plans. Some find that "the likelihood of replacing a founder with a professional CEO more than doubles when VCs first finance the company."
One paper finds no evidence that VCs disguise failures as acquisitions, and another finds that IPOs and acquisitions are highly correlated with higher investor returns. VC-backed companies have been found to be more likely to go public, more likely to be acquired, and less likely to fail, relative to non-VC control groups. "This basic pattern has generally been found in various subsamples over time."
"Overall we would say that the literature provides clear evidence that VC-backed companies achieve better exits than most other start-up categories, and that there is a correlation between certain investor and company characteristics and these exit outcomes. What remains a significant challenge is attributing causal effects to any of these investor and company characteristics. ... Yet by far the biggest knowledge gap concerns the reasons for, timing of, and dynamics around company failures."
A sample of IPOs from 2001 to 2007 found that two-thirds of them had angel investors.
"Overall we woudl say that while there is a substantial body of research that examines the role of VCs in the process of going public and the long-run performance of VC-backed IPOs, there are relatively few results that remain valid across time and geography."
Regarding government venture funds (GVCs), the authors say that the weight of the studies implies "that simplistic judgments about GVCs being either better or worse than private VCs are inappropriate."
Specialized VC firms outperform generalist VCs.
Internal rate of return (IRR) is a popular yardstick, but "several shortcomings make its use in VC problematic."
"There appears to be agreement among researchers that VC returns are not as high as those reported by industry participants and associations. ... Beyond that, however, there is also a lively debate about what the true returns might be." Studies consistently find an average public market equivalent (PME) ratio of around 1.0, indicating roughly equivalent performance with funds invested in the public market. They also consistently find a very wide dispersion of returns.
"Overall we note that while different studies obtain somewhat different estimates of the net returns, there is an emerging consensus that average returns of VC funds do not exceed market returns. Moreover, there is considerable dispersion and skew. While the net returns of the best VC funds are clearly very high, the median VC fund rarely beats the market, and the lower tail of the distribution can generate large negative returns."
"Overall we would say that these are still early days in terms of understanding and measuring the risk and liquidity properties of VC funds, let alone the heterogeneity among different types of VC firms in this respect."
Skill v. luck: some studies have found performance persistence among VC funds. Other studies have found either (a) weaker persistence, (b) no evidence for performance persistence, or (c) performance persistence for the lowest-performing VC funds.
Returns for limited partners (LPs) vary widely. This is related to the type of LP (endowment, pension fund, etc), their organizational structure (investment requirements), and other characteristics.
Some researchers have taken the opposite approach, trying to estimate the entrepreneur's returns: "Three out of four venture-backed entrepreneurs have no returns at all, but the successful ones receive on average $5.8 million at exit. Making several assumptions about the entrepreneurs and their degree of risk-aversion, they argue that the expected utility of being a venture-backed entrepreneur is surprisingly low."
Companies backed by "failure-tolerant VCs are more innovative."
"Overall we believe that the body of empirical evidence is consistent with the notion that VCs select more innovative comapnies, and then help them with the commercialization process. The results suggest that VC plays a greater role for commercialization (as measured by bringing products to market, and forging strategic alliances) than for the generation of further innovation (as measured by patents and [total factor productivity])."
Most studies at the metropolitan and state level find positive correlations between VC, new firm entry, and employment. Company-level studies confirm this, with VC-backed companies exhibiting faster growth both before and after receiving VC investment. The "social costs" of VC are an open question, especially since "the private returns to VC are often disappointing, implying that VC investments have significant opportunity cost."
In terms of policy, studies find that capital gains tax rates affect the supply of entrepreneurial activity and thus, through the demand rather than supply side, the level of VC. Likewise, labor market rigidities reduce VC investments in high-tech sectors. A difference is found, however, between labor protection (trying to prevent layoffs) and labor insurance. The former is associated with lower VC levels.
The Small Business Innovation Research (SBIR) program has been found to be associated with higher employment and sales growth at awardee companies, as well as more follow-on VC funding--though only in certain regions and sectors.
Future: "Some of the areas where more research is needed are the early (pre-VC) history of VC-backed companies, and the choice between alternative sources of financing. Few papers so far shed light onto the internal workings of VC firms ... Much of the work so far, both empirical and theoretical, has been on the VC-company relationship, and only recently have researchers delved into the details of the relationship between GPs and LPs."