Originally published July – August 2018
When managers and scholars talk about diversity’s impact on organizations and teams, they’re usually referring to the effects on collective accuracy and objectivity, analytical thinking, and innovativeness. On “harder” measures of financial performance, researchers have struggled to establish a causal relationship with diversity—particularly when studying large companies, where decision rights and incentives can be murky, and the effects of any given choice on, say, profits or market share can be nearly impossible to pin down.
So we’ve zeroed in on diversity’s effects in the venture capital industry, which presents fewer barriers to understanding. VC firms are fairly flat in structure, composed primarily of investment partners and relatively few junior professionals. Every investor is a decision maker, and choices have clear business consequences. We know which firms make what investments, and for the most part we can identify the individuals leading those investments, because they usually take seats on the boards of portfolio companies. Using publicly available information, we can analyze VC professionals’ “endowed traits,” such as gender and ethnicity, and “acquired traits,” such as schooling and work history. In other words, we can see how similar or different these decision makers are and compare the quality of their decisions on the basis of their investments’ performance. Because their incentives are aligned and readily discernible—compensation for VCs is largely determined by profit sharing, ensuring that they and their investment partners have the same goals—the analysis is not clouded by inscrutable interests. The goal of every venture capital investor and firm is to choose and groom the companies that will yield the best possible outcomes.
All in all, we couldn’t have asked for a better “lab rat” than the VC world. Over the past several years one of us (Paul Gompers) has examined the decisions of thousands of venture capitalists and tens of thousands of investments, and the evidence is clear: Diversity significantly improves financial performance on measures such as profitable investments at the individual portfolio-company level and overall fund returns. And even though the desire to associate with similar people—a tendency academics call homophily—can bring social benefits to those who exhibit it, including a sense of shared culture and belonging, it can also lead investors and firms to leave a lot of money on the table.
In this article we’ll describe the research behind those findings and provide recommendations for reaping the business benefits of diversity. Decision makers fare best when they openly acknowledge and address homophily early on, understand that small adjustments in mindset and behavior can have lasting ripple effects, and diversify their personal as well as professional networks.
The gender and racial makeup of the venture capital industry is staggeringly homogeneous. A comprehensive data set of every VC organization and investor in the United States since 1990 shows that the industry has remained relatively uniform for the past 28 years. Only 8% of the investors are women. Racial minorities are also underrepresented—about 2% of VC investors are Hispanic, and fewer than 1% are black. Those groups have seen significantly increased representation in other fields and in advanced professional and scientific degree programs, but not in the VC industry. It’s against that backdrop that venture capitalists choose their collaborators at other firms, investing their money side by side and joining the boards that guide the start-ups. Most investors specialize in a particular industry or sector, so potential partners are easy for researchers like us to identify: They are investing in the same types of deals at around the same time. And venture capitalists are far more likely to partner with people if they share their gender or race. They’re also significantly more likely to collaborate with people if they share their educational background or a previous employer. Belonging to the same racial group increases the propensity to work together by 39.2%, and having a degree from the same school increases it by 34.4%. Not only is the likelihood of collaborating on any one deal greater, but VCs tend to keep teaming up with those who share their traits.
What does all that mean for performance? How do the financial outcomes of homogeneous partnerships compare with those of diverse collaborations? The difference is dramatic. Along all dimensions measured, the more similar the investment partners, the lower their investments’ performance. For example, the success rate of acquisitions and IPOs was 11.5% lower, on average, for investments by partners with shared school backgrounds than for those by partners from different schools. The effect of shared ethnicity was even stronger, reducing an investment’s comparative success rate by 26.4% to 32.2%.
To understand why homogeneous teams have worse investment outcomes, it’s critical to determine exactly when decision making suffers. Interestingly, projects selected by both homogeneous and diverse sets of investment partners were equally promising at the time the decision to invest was made. Differences in decision quality and performance came later, when the investors helped shape strategy, recruitment, and other efforts critical to a young company’s survival and growth. Thriving in a highly uncertain competitive environment requires creative thinking in those areas, and the diverse collaborators were better equipped to deliver it.
Of course, the industry’s homogeneity is continually reinforced by individual firms’ hiring decisions. Because these organizations are small (they usually have three to five investment professionals), and spots open up infrequently (every two to four years), even a slight preference for candidates who are similar to existing partners has a lasting effect. Here’s just one example: Many prominent venture capital firms were founded by Harvard Business School alumni, and now nearly a quarter of all VCs with MBAs come from Harvard. To put that into perspective, only 9% of VCs with MBAs are from Wharton, and just 11% are from Stanford—both top-tier schools.
Prospects are even worse for female candidates. Remember that only 8% of venture capital investors are women. It’s no wonder, since nearly three-quarters of VC firms have never hired a woman in that role. What separates that overwhelming majority from the firms that have hired women? One powerful factor is the gender of the partners’ children. When a firm’s partners have a higher proportion of daughters, the likelihood that a female investor will be hired goes up significantly. Simply replacing one son with a daughter would increase the probability of hiring a woman by 25%.
Only 8% of VC investors are women. Fewer than 1% are black.
Of course, we aren’t suggesting that male VCs should have daughters to reduce gender bias and increase diversity in their firms. But because the gender of one’s child isn’t a choice, the finding offers a tighter lens on diversity’s effects. When the “daughter effect” does bring more women into the fold, it has a strong impact on performance. Venture capital firms that increased their proportion of female partner hires by 10% saw, on average, a 1.5% spike in overall fund returns each year and had 9.7% more profitable exits (an impressive figure given that only 28.8% of all VC investments have a profitable exit).
The economic impact of diversity isn’t limited to the VC world. A recent NBER analysis of highly skilled occupations (in fields such as law, medicine, science, academia, and management) shows a positive relationship between diversity and the value of goods and services produced in the United States. The study looks at GDP trends beginning in 1960, when significant barriers prevented white women, black women, and black men from entering those professions. Though we’re still nowhere near parity, gender and racial diversity have increased markedly in such fields over the past 50 years—and the U.S. economy has grown in that same period. Using a model that assumes innate skills are evenly distributed across gender and racial groups, the NBER analysis attributes about 25% of the GDP growth per capita to the uptick in white women and black Americans of both genders. In short, the authors argue, the United States began making better use of the talent at its disposal.
Given that homogeneity imposes financial costs and diversity produces financial gains, a natural next step is to assess what managers can do to increase representation across groups. Here are some evidence-based recommendations:
Timing is a crucial and often overlooked factor. Founders and entrepreneurs in particular may place diversity low on their list of early priorities, viewing it as a concern that can be addressed once their firms have grown. But it is far easier to build a diverse organization from the ground up than to diversify a large, complex, homogeneous machine.
Stacy Brown-Philpot, the CEO of the freelance-job site TaskRabbit, made that point when she reflected on her early days as a financial director at Google. “When I joined Google, it was 1,000 people,” she said. “It took me two and a half years to look around and realize there weren’t a lot of people like me. So [my colleague] David Drummond and I…put together a group. It was really late. I think that’s part of the challenge [at Google].” When Brown-Philpot moved to TaskRabbit, she took a different tack with the young company, partnering with the Congressional Black Caucus’s CBC TECH 2020 initiative to bring more black workers into the tech industry. In 2016 Brown-Philpot publicly committed to increasing TaskRabbit’s black workforce from 11% to 13% of employees by the year’s end, to ensure proportional black representation at the company.
Sociology scholarship underscores the flaws in a delayed approach. In one study researchers used e-mail as a proxy for social connections at a university. They discovered that over multiple “generations” of interaction, such as taking new classes or joining new activities, even minor individual tendencies to interact with similar people could have a large cumulative effect, resulting in striking levels of group homogeneity. The result suggests that an already homogeneous organization will tend to become even more so as it scales up. So it’s important to encode diversity in a company’s DNA at the earliest stages.
This is not to say, of course, that it’s impossible to improve diversity in an established company. Standardized processes, such as blinding résumés during hiring and using objective metrics during performance reviews (as long as they’re constantly refined through iterative development), can have a big impact in organizations looking to ameliorate bias. But when the teams developing and refining those processes are themselves unrepresentative of the broader universe of candidates, they must take special care to ensure that they aren’t institutionalizing their individual biases.
This is true not just in venture capital and entrepreneurship but in any setting where small groups of people wield outsize decision-making authority. Bringing just a few talented women or racial minorities into a group like that changes the relative balance of power. And recent findings suggest that if those individuals make hiring decisions, they will affect the group’s future makeup. In an online simulation, participants were placed in “employer” and “potential hire” buckets. Choosing between one woman and one man, female employers hired the woman 50% of the time, while men hired her only 40% of the time.
That might be interpreted as evidence of affinity, suggesting that the homophilic biases that can hamper diversity when exhibited by overrepresented groups can bolster it when exhibited by underrepresented ones. Or the results might suggest that people who have been historically disadvantaged in recruiting are less likely to discriminate against those who share their endowed traits. Both explanations are probably true to some extent. But one of us (Gompers) actually found in a recent study that members of traditionally underrepresented groups were more likely than white men to seek out people unlike themselves when forming entrepreneurial teams. That result implies that qualified members of dominant groups aren’t in much danger of being locked out of diverse organizations. Combined with the fact that group homophily tends to compound over time, it also suggests that if the goal is proportional representation over the long term, it’s better to overcorrect for bias early on, by hiring more people from traditionally underrepresented groups, than it is to undercorrect.
To accomplish that, companies need not explicitly favor a particular race or gender when hiring. Sometimes simple adjustments in the selection process can increase diversity. In one study led by the behavioral economist Iris Bohnet, of Harvard Kennedy School, students were assigned the role of an employer asked to select an employee who would do well on a future math or verbal task. Even though gender was not predictive of performance, “employers” evaluating individual candidates were likely to be swayed by stereotypes, exhibiting a preference for women on verbal tasks and men on math tasks. But when they assessed two candidates side by side, gender suddenly became irrelevant. Evaluators instead focused on past performance—an actual indicator of future success.
We’ve seen similar results in blind evaluations of prospective hires. Most of us have heard that auditioning musicians behind screens has dramatically increased the percentage of women who make the cut for symphony orchestras. Here’s an example from another industry: When the political satire show Full Frontal with Samantha Bee was gearing up to hire writers, then-showrunner Jo Miller combined other shows’ evaluation processes, making minor tweaks consistent with her goals. In a first-round call for script submissions, detailed formatting instructions were included so that superficial indicators of experience would not overshadow talent, taste, and potential. Those scripts were evaluated blindly, and an unusually large number of applicants made it to a second round, in which previous work and other factors, including gender and ethnicity, were considered. The result was a strikingly diverse team for late-night comedy: 50% women and 30% people of color.
Though these were basic process adjustments, another important ingredient is intention. Both Miller and Bee felt that a diverse writers’ room was a priority for the show, given its subject matter and irreverence. The hiring process was deliberately designed to support that goal. But that’s not the case in most organizations.
Consider the typical newsroom. The American Society of News Editors’ 2017 Newsroom Employment Diversity Survey found that almost every major newspaper in the nation, from the New York Times to the Boston Globe to the Washington Post, is whiter than its audience city. When the New York Times Magazine reporter and 2017 MacArthur fellow Nikole Hannah-Jones was asked to offer advice to journalists of color in light of the troubling report, she instead issued a call to newsroom managers to examine whether “their stated goals are really their goals.” She added: “If newsroom managers wanted diverse newsrooms, they’d have diverse newsrooms.”
Other prominent figures in the media shared this assessment. The New York Times columnist Charles Blow reflected in a recent tweet, “As a newsroom manager from age 25 to 37, [I] was always struck by how the ‘soft skills’ [people] favored were in many ways culturally exclusive.” The broadcast journalist and producer Soledad O’Brien passionately concurred. “It is not brain surgery,” she noted.
Because social and professional circles often overlap, homogeneous personal networks can have a deleterious effect on organizational diversity. That’s why some companies have deemphasized referrals, or at least cautioned against their pitfalls. But reliance on personal networking is still crucial to the functioning of certain industries. A survey of venture capitalists, for example, showed that social connections are essential to generating deal flow. But investors’ personal networks tend to be closed, given that most VCs have the same educational background, are the same gender and race, and have worked at similar firms. Consequently, they can miss a lot of opportunities.
Though assigned mentorship and other professional programs can help decrease bias and increase diversity in organizations by exposing managers and employees to more people who are less like them, such relationships are by nature hierarchical and may actually aggravate individuals’ prejudices. In one study, when white participants were assigned the role of “superior” over a black subordinate, their racial bias increased. Situational power in same-race pairs had no impact on racial attitudes.
At the individual level, extensive social contact on an equal footing is a better strategy for lessening bias. One representative study demonstrated that friendships with homosexual individuals were effective in reducing sexual prejudice. Another study found that white participants’ friendships with Latinos or African-Americans reduced their implicit biases toward those groups.
The most generous interpretation of homophilic tendencies is that they arise from a seemingly innocuous desire to interact with people like ourselves. But the analysis of entrepreneurial team formation mentioned earlier revealed that endowed traits had a stronger homophilic “pull” than acquired traits. Social interactions can compel people to reevaluate what it means for someone to be “like them,” beyond such easily discernible demographic indicators. The benefits of these interactions carry over to the workplace, where expanded networks and mindsets can improve both individual and organizational performance.
A willingness to openly recognize and tackle bias is at the heart of all our recommendations. When people choose to ignore bias or deny that it exists, they keep seeking out business partners, team members, and employees who share their traits, and they miss out on the quantifiable benefits of diversity.
Social science research suggests that people tend to react with anger and irritation when confronted about their biases—particularly when those biases are accurately labeled as such. Although such interactions may be unpleasant, they also tend to lead to behavioral change, and so should be welcomed as opportunities for growth. Bias is a measurable condition, but it is not a permanent one, on either the individual or the organizational level. By acknowledging it we can counter it, expanding our pool of potential collaborators and improving financial performance.