The Handoff: When a New Firm Leaves the University’s Orbit
For an academic spin-off, incorporation answers only one question: has a company been formed? It does not answer whether the company is active, surviving, selling, or growing. This chapter focuses on the transition from launch to sustained operation. By separating those outcomes and following them across three-, six-, and nine-year horizons, we can see why the conditions that help a university create a spin-off are not automatically the conditions that help the spin-off expand.
Growth is a different variable from launch
The longitudinal Italian study measures the two stages differently because they represent different problems. For the launch stage, the outcome is the number of new academic spin-offs per 1,000 academic researchers. This asks how strongly a university setting is associated with the formation of new ventures relative to its research population. It is a measure of venture creation, not a measure of what happens after the legal entity exists.
For the growth stage, the study looks at active spin-offs’ sales three, six, and nine years after establishment. Its main outcome is a yes-or-no indicator: is an active firm’s sales above the median value at that horizon? The researchers also test mean sales as a robustness check. That design has two important consequences. First, “growth” means relative commercial performance within the observed group, not a universal revenue threshold. Second, the denominator is active firms with available sales information, not every company that was ever launched.
The growth analysis excludes companies established in 2017 and 2018 because financial information was not yet available three years after their creation. The total group used for the growth analysis is therefore 822 ASOs, not the full launch sample of 945. That difference is not a defect to hide; it is part of what the measure means. A three-year sales comparison cannot include a firm that has not yet had three years to produce the relevant observation. The same logic applies when comparing six- and nine-year outcomes: the population becomes smaller and more selected as the required observation window lengthens.
“Active” also needs care. A firm can remain open while making little progress, and a firm can have modest sales while performing valuable specialised work. Closure is not the same as commercial failure, just as survival is not the same as growth. Sales are useful because they reveal whether the venture is converting its technical or research base into market activity, but they do not capture every form of innovation, employment, social value, or research continuity.
The difference between the two measures can be stated simply. Launch asks whether a research environment is producing new companies. Growth asks whether companies that remain observable are converting their initial position into continuing economic activity. The first is close to the university’s formation machinery. The second is exposed to customers, capital, institutions, talent, and the firm’s own ability to execute.
The lifecycle contains both transitions and filters. A firm must remain active to appear in later observations, and it must have sales data to be compared. Those filters are analytically necessary, but they mean that later results describe a selected population. A nine-year pattern among surviving, observable firms cannot be read as the destiny of every company launched nine years earlier.
Consider three illustrative firms in the same launch cohort. One closes before the first sales observation and disappears from the active-firm comparison. A second remains open but records sales below the median, so it is counted as active but not as above-median growth. A third exceeds the median and remains visible through the later horizons. A launch statistic sees all three as companies created. A growth statistic sees them through different filters and asks a narrower question about relative sales. None of these measures alone tells the whole story, but confusing them would make the research appear to answer a question it was not designed to answer.
The median threshold is also a comparison device, not a claim about commercial adequacy. A firm can be above the median while still having modest sales, or below the median while developing an important technology whose market takes longer to mature. The measure is valuable because it can compare firms at the same horizon and reveal which institutional or firm characteristics are associated with relative performance. It is not a substitute for asking whether the sales level covers costs, supports hiring, or creates durable value.
What the numbers say about an uneven transition
The descriptive results make the distance between launch and growth concrete. Among active ASOs observed three years after launch, about 58 percent had sales below €50,000, while fewer than 2.5 percent had sales above €500,000. Median sales were approximately €32,000 at three years, €62,000 at six years, and €88,000 at nine years. Mean sales were much higher—about €92,000, €173,000, and €290,000 at the same horizons—because a relatively small group reached much larger values.
The gap between the mean and median is not a statistical footnote. It describes the shape of the population. The median tells us what happened around the middle of the observed firms. The mean is pulled upward by the companies that sell much more than most others. If a policymaker looks only at the mean, a small number of high performers can make the typical spin-off appear further along than it is. If the policymaker looks only at the median, they may miss the conditions that allow a minority to break out. Both measures are useful, but they answer different questions.
The same distinction applies to company counts. A launch count measures the output of a formation system. It can reveal whether researchers are forming ventures and whether a university has developed channels for doing so. It cannot show whether those ventures are finding customers or becoming productive employers. The study reports that 143 ASOs ceased activities between 2004 and 2018, about 17.3 percent of the total sample. That is a descriptive observation for this Italian sample and period, not a general closure rate for academic ventures everywhere. Even among firms that did not close, the sales distribution shows that activity and growth were far from equivalent.
Time changes the question again. At three years, an ASO may still be converting a prototype into a reliable offering, negotiating its first contracts, or using a university’s incubation conditions. At six years, it may need repeatable sales, a larger team, and working capital. At nine years, the constraints may include physical and digital infrastructure, access to finance, market density, and the ability to keep innovating after the original research advantage has been absorbed. These are not fixed stages with universal deadlines. They are different kinds of exposure to the outside world.
What persists across the transition is the technical and organisational base the university helped assemble: the knowledge, rights, founder relationships, and initial credibility may continue to matter. What can be lost is the protected setting around them. A university laboratory may absorb costs that a stand-alone firm must later pay. An academic relationship may open a first door without supplying a repeatable distribution channel. A founder who can lead research may still need a commercial or operational partner. Growth is the point at which these hidden dependencies become visible because the firm must reproduce its performance outside the original support environment.
This is why the sales horizon should be read alongside the mechanism behind it. If a firm falls below the median at three years, the reason could be weak demand, delayed validation, inadequate capital, or a rule that slowed a necessary transaction. If it remains below at nine years, the relevant explanation may include persistent infrastructure, finance, talent, or market-access constraints—but the dataset cannot identify every one of those paths for each firm. The horizon tells us when an outcome is observed; it does not, by itself, tell us which intervention will fix it.
The regression results follow that changing exposure. The aggregate Institutional Quality Index has a positive, statistically significant association with growth at three and six years, but not at nine years. This does not show that institutional quality stops mattering. It shows that the aggregate measure was not statistically significant in that specification at the later horizon. The article suggests that early growth may depend more on interaction with local institutions, while longer-term development is affected by other features of the local context.
At nine years, physical and digital infrastructure and easier access to finance become positive and statistically significant in the study’s discussion. Operating in a more densely populated province, though not simply in a province with a large total population, is associated with growth, and being innovative also matters. The shift is plausible as a lifecycle mechanism. Early on, a young firm may be most vulnerable to unclear rules, slow decisions, or weak administrative credibility. Later, even a well-governed environment cannot make up for a shortage of capital, customers, skilled collaborators, reliable connectivity, or facilities needed to deliver at scale.
This interpretation also clarifies why a long-term comparison cannot be reduced to “the same support, but more of it.” A university may help a founding team protect an asset, build a prototype, or reach its first partner. Those contributions can be decisive at launch. They do not eliminate the need for external finance or a market in which the company can repeat its transactions. A support instrument can work at one transition and fail at the next because the firm has acquired a new dependency.
Selection makes the later results harder, not less, interesting. Firms observed at nine years are not a random sample of all firms created at the beginning of the period. Some have closed. Some may have become inactive or difficult to measure. Some may have remained small but stable. The surviving group may have different technologies, founders, ownership structures, and access to capital than the original group. A positive or non-significant association at nine years therefore describes the firms and measurements that remain in the model; it does not settle what would have happened to firms that disappeared from view.
Sales also carry a boundary that matters for policy. A low-sales ASO may be underperforming, but it may also be pursuing a slow-moving technology or serving a narrow market. A high-sales firm may be commercially successful without creating the broader spillovers policymakers expect. The study’s sales measure is useful for exposing the difference between formation and expansion, but a balanced evaluation would place it beside survival, innovation, employment, research continuity, and the quality of external finance. None of those measures should be used to erase the others.
The practical lesson is to diagnose the stage before choosing the intervention. If a new firm cannot clarify its rights or validate its first proposition, the relevant support may still be close to the university. If it has a credible product but cannot obtain working capital, recruit, reach customers, or operate through slow public systems, the leading constraint lies beyond the original launch platform. The same company can move from one category to the other without any contradiction in the evidence.
This is why growth should be treated as a handoff rather than a delayed version of launch. The university may supply the initial technical and organisational bridge. The local ecosystem determines whether the bridge connects to roads that lead to finance, customers, talent, infrastructure, and enforceable agreements. The next chapter examines those external conditions directly, beginning with what institutional quality means in practice and why it matters at different points in an ASO’s growth.