Before the Pitch: Entry Shapes the Network
For the Indian high-growth startups in this study, the important question is not simply how many people a founder can name. It is how the study turns recorded backgrounds and venture roles into evidence of a potential connection, and why that potential is shaped by the way a startup enters the market. Understanding those choices lets us distinguish proximity from activation before asking whether a network becomes capital.
A connection begins as a record, not a conversation
The study uses information about firms and the people associated with them: founders, board members, angel investors, executives, and other team members. It then looks for common ground in recorded education and professional experience. At the broadest level, this produces measures of potential educational and professional connections and counts of matched links. These measures are intended to capture a founder’s position in a wider network of people and institutions.
That is a useful starting point, but it is not the same as observing a relationship in ordinary language. If a founder and an investor attended the same kind of institution, or if their biographies contain overlapping professional experience, the data can register a potential match. It cannot tell us whether they know each other, whether they have exchanged advice, or whether either person would answer the other’s call. The measure identifies a possible reduction in social distance, not a completed introduction.
The study also constructs a narrower set of relationship measures. It separately examines links between a founder and a board member, a founder and an angel investor, and a board member and an angel investor. These are not merely bigger versions of the broad network count. They describe a relationship inferred from people who appear around the same venture and whose recorded education or experience matches under the study’s rules. The distinction is essential: a founder can have substantial potential centrality without appearing in a measured founder–angel link.
A simple illustration makes the difference clear. Suppose a founder’s biography shares an institution or prior employer with an angel investor listed in the startup’s records. The broad measure may treat that overlap as potential educational or professional proximity. The role-specific measure asks a different question about the people attached to that venture. Neither measure observes the conversation in which the founder is introduced, the investor’s assessment of the opportunity, or the terms of a possible investment. The recorded match is evidence about the structure of the network, not a transcript of its use.
This distinction also protects against a subtle form of individual blame. When a potential connection does not become an investment, the missing step may involve recognition, trust, relevance, timing, or the investor’s own constraints. The data do not let us assign responsibility to the founder simply because a match was recorded. Nor do they let us infer that an investor ignored a clearly available opportunity. They show that a possible tie and a consequential tie are different states.
The network therefore has at least two layers. The first is an opportunity structure: who appears reachable through shared education or professional experience. The second is a venture relationship: who appears connected to the founder, board, or angel-investor roles in the firm. A third step—whether that relationship produces sponsorship, participation, or capital—belongs to the funding analysis. Keeping the layers separate is not pedantry. It is the only way to see where conversion may fail.
The role distinctions also carry different possible routes into capital. A founder–angel link is closer to a direct financing relationship; a founder–board link may bring advice, credibility, or access to other decision makers; a board–angel link may connect governance and investment networks. The dataset does not tell us which of these functions occurred, and it does not say that one link is better than another. It simply allows the study to ask whether the people occupying those roles are related in the recorded network. A startup may have all three links, only one, or none, so the indicators should not be read as a ladder that every firm must climb.
The count measures have a similar limitation. A high count can mean that a firm is connected to many people, but it does not show whether those people are influential, available, or relevant to the next financing decision. A single well-placed relationship may matter more than many weak matches, yet the data cannot measure that quality. Conversely, a missing match is not definitive evidence that no relationship exists: a biography may be incomplete, a tie may have formed through family or community rather than a shared institution, or the connection may not be visible in the records. This is why the broad measures are best treated as maps of observed opportunity structure, not inventories of the founder’s social capital.
The diagram deliberately ends before investment. A role-specific link can be closer to a usable relationship than a broad background match, but it is still not proof of an investor’s decision or a capital transfer. In the study’s terms, the network measures are observations built from available firm and biography records. They are not direct measures of trust, reciprocity, introductions, or negotiating power.
Entry changes the field before fundraising
The network a founder can use is partly a product of where and when the startup begins. The World Bank study reports that women face higher entry costs and tend to enter sectors that are high-growth at the time of entry, rather than sectors whose growth has not yet arrived. Its interpretation is that this safer timing can mean foregoing gains from entering a sector before it ripens. Whether the pattern reflects constraint, opportunity set, preference, or some combination, it changes the environment in which later networks are formed.
The reproduced entry models give some shape to that claim without proving a complete explanation. In a specification with state, sector, and cohort fixed effects, the coefficient on prior sector growth is 0.602, while the interaction between sector growth and woman founder is −0.0614 and is not statistically distinguishable from zero. The first estimate says that prior sector growth is strongly associated with the measured entry outcome. The second does not show a reliable gender difference in that association in this model. It would be an error to translate either coefficient into a direct monetary estimate of entry cost.
A separate sector-cohort comparison shows why persistence must be handled carefully. The lagged share of woman-founded firms is strongly associated with the current share when sector fixed effects are absent, with a coefficient of 0.728. Once sector fixed effects are included, the estimate falls to 0.123 and is not statistically significant. Some apparent continuity may therefore reflect stable differences between sectors rather than a general process that reproduces women’s participation everywhere. This is an important lesson in reading fixed effects. Adding sector effects asks the model to compare firms while accounting for persistent differences between sectors; it does not recreate the missing history of how a founder chose a sector or what alternatives were available. A coefficient that changes after this adjustment is not a failed result. It tells us that part of the raw relationship was entangled with where firms were founded, while the remaining estimate still needs an interpretation suited to the data.
For network conversion, the distinction between sector-level opportunity and founder-level choice is practical. An emerging sector may contain fewer established board members or angels, so a founder’s potential network may be broad in the abstract but thin in the specific roles needed for financing. A mature sector may offer many investors but also a crowded set of recognizable credentials. Neither environment is automatically better. The relevant question is whether the founder can turn the available relationships into evidence that matters for the venture at its current stage. The study’s entry results point toward this dependency; they do not measure the quality of each sector’s introductions or the terms of each financing conversation.
These results matter for networks because sectors are not neutral containers. A founder entering financial technology encounters a different set of investors, professional communities, technical expectations, and visible success stories than a founder entering a consumer or manufacturing sector. The evidence does not measure every one of those interactions, but the mechanism is straightforward: sector choice changes which people are relevant and which forms of experience are legible. The same number of potential contacts can have different value when they are connected to the startup’s actual market and financing stage. A contact who is helpful during product validation may not be the person who can underwrite a later round. An experienced operator may be able to open a door but not provide capital; an angel may provide early money but not the institutional credibility needed for scaling. These are plausible distinctions in the startup process, not separate effects estimated by the entry table. They show why the network question should always be paired with a stage, a role, and an outcome.
The timing trade-off is equally important. Entering a currently fast-growing sector can make demand and investor attention easier to explain. It can also expose a startup to dense competition, crowded narratives, and investors who compare many similar opportunities. Entering before a sector ripens may require more patient capital and more explanation, but may offer room to define a category. The World Bank abstract supports the claim that women founders tend to avoid some of those not-yet-ripe opportunities; it does not establish whether the choice is voluntary, forced by entry barriers, or economically optimal for any particular founder.
That uncertainty should change how the result is used. It is not a prescription that women founders should enter earlier, choose riskier sectors, or disregard current demand. Nor is it evidence that a “safe” sector choice is irrational. It is a warning that the apparent network available to a founder is conditional on the route into the ecosystem. A strategy that works in a mature, highly visible sector may fail in an emerging one because investors, validators, and potential board members are distributed differently. Conversely, a network built in one sector may not transfer cleanly when the startup changes its market.
Where the measurement stops
The study’s network construction has a clear strength: it makes otherwise invisible social structure comparable across many firms. It can count education and experience matches and distinguish founder, board, and angel roles at a scale that interviews alone could not easily provide. It also makes the central puzzle testable. We can ask whether women founders show more potential centrality, then ask whether that pattern is reflected in the role relationships around their ventures.
But the same construction sets a boundary around the conclusion. A matched biography can be a weak proxy for a live relationship. A missing match can mean no shared background, incomplete biography information, or a relationship formed through another route. The public package does not observe whether a tie is reciprocal, whether one person has more status than the other, or whether a founder can activate the connection at the moment capital is needed. These are not small details: they are precisely the conditions that determine whether proximity becomes sponsorship. Stronger evidence about conversion would need to follow relationships over time, record actual introductions and rejected pitches, or combine the firm database with interviews or survey measures of trust, reciprocity, and investor evaluation. It would also need to distinguish an absent relationship from a relationship that is simply missing from an online biography. Until then, the most responsible use of the network variables is comparative: they identify where the observed structure differs and where a more direct explanation is still required.
Entry conditions add another boundary. The regressions control for observable features such as sector, state, cohort, firm age, and city tier in the specifications reported here, but those controls do not reconstruct the founder’s full opportunity set. They cannot reveal every resource available at founding, every constraint on time or mobility, or the quality of the market opportunity that an investor perceived. A result that survives fixed effects is more informative than an unadjusted comparison, but it remains a conditional association.
This leaves us with a disciplined starting point. The study does not ask whether a woman founder possesses the right personality for networking. It asks how observed firms differ in the network structures that surround them, and how those structures relate to entry and later outcomes. The next question is therefore empirical and specific: when the broad measures show potential connections, do women founders actually have more of them—and do those potential connections resemble the role-specific relationships through which a venture might gain access to capital?