The employment destination of PhD-holders in Italy: non-academic funded projects as drivers of successful segmentation

In all developed countries in recent years, the non-academic labour market destination of PhDholders (segmentation) has emerged as an issue. Universities and other research-intensive institutions can no longer absorb the major share of PhD-holders. Their employment has become a matter of segmentation both in horizontal (economic sector) and in vertical (income) dimensions. The paper tests what factors segment labour market outcomes in both dimensions – economic sector and income. Findings suggest that whilst scientific mobility and type of funding during PhD period do not play a notable role, some types of experiences such as post-doc, instead, predict exit from academic employment and also a higher income overall. The most significant experiences that contribute to segmentation are in fact projects funded by private companies or international organizations in postdoctoral periods. Policy-making implications are relevant for both PhD-holders, universities and external organizations. For instance, non-academic employers may maximise their collaborations with universities with beneficial outcomes for PhD-holders themselves as well


Introduction
A wide body of literature has pointed out that the labour market of PhD-holders, especially the non-academic one (Mangematin 2000;Enders 2002, Caparros-Ruiz 2019Cruz-Castro, & Sanz-Menéndez 2005;Jackson & Michelson 2015;Neumann & Tan 2011, Pedersen 2016van der Weijden et al. 2016), has become critical. PhD-holders have been detected as a target of people under segmentationmeaning that PhD-holders will have to get non-academic positions in any economic sector due to shortage of positions in academia and in other public research institutes, especially tenured or secure positions. Nevertheless, the body of knowledge about what may facilitate segmentation is poor in relation to this increasing phenomenon. What does fully unleash the potential of PhD-holders when they can't or don't wish to continue a career in academia? This problem is increased by the evidence that the arguable attractiveness of such highly qualified education does not necessarily occur spontaneously by non-academic employers (Casey 2009). Hence, to understand the drivers for such not yet fully developed segmentation is important. The topic is intriguing because arguably PhD-holders may prefer to work as researchers although opportunities for permanent positions are poorer in comparison to some decades ago (van der Weijden et al. 2016;Haley et al. 2018). In analysing a context like PhD-holders from Italian universities, the paper explores one of the most difficult OECD country in which to pursue an academic career, exacerbating the pressing issue of getting alternative employment. The paper investigates which steps in a PhD-holder's career facilitate a transition towards non-academic and non-public research institute careers (this representing the first hypothesis) and which steps make one's career a successful onemeaning resulting in having higher income, the second hypothesis. Push-factors in contemporary academic labour market are the background theory that generates these two hypotheses regarding sector destination and differentials in wages. By means of using a dedicated survey about PhD-holders employment destination after several years of PhD attainment, the paper contributes to this debate finding that PhD-holders are more likely to get positions outside academia when they worked in projects funded by non-academic organisations. Non-academic post-doctoral projects are also prediction of establishing oneself as self-employed, giving further insight and wider perspective to the topic of academic spin-offs. This relationship between PhD-holders and non-academic organisations via post-doctoral projects that took place in universities is also a successful gateway for higher incomes. In order to deepen knowledge, we conceptually assume that multiple events across a career trajectory may constitute patterns of labour market destinations. To take into account this well-known dynamic in life cycles, the paper provides extensive endogeneity analyses using key past events about doctoral and postdoctoral experiences as instrumental variables. The paper is organised in the following way. The literature review outlines the main debates in the field of PhD career trajectories. The Data section describes the dataset and its variables used for the analysis. Hypotheses and results section provide the two hypotheses: 1) what does predict a non-academic employment; 2) what does predict a higher salary. A subsection provides evidence of tests of endogeneity in order to check whether main events sorted chronologically determine following career steps. The Discussion section supplies more in depth implications with literature. Conclusions attempt recommendations to policy makers at European level also out of the specific Italian higher education context.

Literature review
The topic of employment of doctoral holders is relatively recent in Europe or other developed countries. When facing this topic and looking primarily on destination employment, some clusters of relatively homogenous bodies of research emerge. These clusters identify following subsections: involvement in projects and the relevance in a timeline; the spin-offs option; the appreciation of the PhD title; the post-doc condition per se; discriminations by gender; factors facilitating job matching; the realisation of one's chances to be academics forever. This section also provides some details of the Italian context in the larger European sphere, which is propaedeutic for the interpretation of these results in a broader European perspective.

Involvement in post-doctoral projects.
Projects are the main mode of work at doctoral and post-doctoral levels. Projects may differ by their nature: some are publicly funded for unfettered research; some others aim at facing a socio-economic problem within one or more disciplines. The latter typology is arguably the most compelling for non-academic employment development. Non-academic organizations, for instance, may expect from PhD-holders that they bring collaborations with universities along with research, development activities and functions (e.g. R&D capacities), plus other skills (Manathunga et al. 2009). Nevertheless, an ex-post study among non-academic employers found that only around a fifth of them detects specific reasons for having hired such a highly qualified personnel (Haapakorpi 2015). Following this evidence, it is reasonable to investigate patterns of over-education and/or over-skilling (Gaeta et al. 2017). Involvement in different post-doctoral projects arguably may predict different eventual career steps. For instance, Garcia-Quevedo et al. (2012) analysed PhD-holders destinations in depth, describing the opportunity to have non-academic employmentnot necessarily private companiesby previous events. In addition, the perspective of different types of projects carried out as a researcher is relevant though unexpectedly unexplored in this topic. A project funded by a company for a scholarship of a PhD student or for a post-doc is a joint investment in a specific person holding the post. A PhD student and a post-doc researcher, though costing in some cases almost the same according to Italian legal framework, might display considerable differences in terms of ability to produce results. This is a cogent point from the employers' perspective, if one follows a human capital assumption. Nevertheless, this issue appears underdeveloped in literature.

Spin-offs.
Spin-offs are one interesting alternative to traditional academic jobs, preserving R&D intensity and exploiting PhD-holders' expertise Meoli et al. 2018). The main difference between these studies and the findings from this analysis is that self-employment as a category includes a larger set of workers in comparison to spin-offs. Self-employment is also not necessarily happening in campuses, in specific labs, or under specific policies, possibly aimed at tackling paucity of opportunities (Rizzo 2015).
Appreciation of PhD title. The title of PhD per se may change by country. Also the structure of the labour market in a country as a whole is important. The Austrian labour market for PhD-holders is very different from the Italian one, for instance: in Austria recent cohorts of PhD-holders are more likely to work in companies and much less likely to do work in education sector other than higher education one (Schwabe 2011). In Italy education and training other than higher education remains a sort of second-best option with lower wages, but also showing less gumption, such as strategic use of social networks (Baruffaldi et al. 2017). Some other specific features by country are worth mentioning. Literature highlights the importance of connecting doctoral programs to destination employment in Anglo-Saxon countries (Rominger 2018), or more broadly in OECD ones (Gokhberg et al. 2016). This aspect is less developed in Italy. Last, in Italy the Anglophone expression PhD is mostly unknown outside academia, and the Italian "Dr." identifies Master graduates also, generating confusion in acknowledging the possible surplus of competence by PhD-holders.
The post-doc condition. A large corpus of research deals with post-doctoral experiences. Some previous studies highlight the increasing number of fixed-term post-doc positions within research-intensive universities, finding that post-doc period(s), if not too long in duration, are somehow beneficial regardless (or with uncertain relation with) prestige of hosting institutions (Su 2013). However, other evidence reveals that postdoc positions are potentially endless (Powell 2015). Other authors argue that the post-doctoral condition itself is pernicious inasmuch uncertainty per se is detrimental (Signoret et al. 2019). In a not very stratified higher education system like the Italian one, it is likely that post-docs positions constitute a sort of "survival tournament" where some PhD-holders get a tenure as soon as they can, while others may drop towards other non-academic or non-public scientific positions. This may happen in an unpredictable mode and with unclear patterns (Wöhrer 2014). To this regard, it is relevant to understand the role of these post-doctoral positions to determine the extent to which different post-docs and non-tenured experiences predict any exit from academic positions. On top of this point, postdoctoral duration is arguably subject to the rule of diminishing returns (Su 2013;Webber & González Canché 2015). At the same time, the higher the scientific productivity, the higher the probabilities to continue to progress in universities or public research institutes, according to new career progression patterns based on scientific production (Marini 2017). All in all, it is unclear the extent to which post-doctoral positions are predominantly positive or negative experiences.

PhD-holders and gender.
PhD-holders are not immune to gender discrimination. Gender pay gap among PhD-holders has been found (Platow 2012;Schwabe 2011;Webber & González Canché 2015), although people in higher percentiles in earnings do not suffer this disequilibrium (Canal-Domínguez & Wall 2014). This latter evidence leaves room for speculations about possible family burdens as a concurring factor of the gender pay gap.
Factors facilitating job matching. The public debate and the discourse about the role of PhD-holders outside academia in Italy is not systemically covered by any support organization, like for instance Vitae in the UK. There is also poor attention towards what a PhD student ought to develop in order to increase his/her probabilities to get valuable positions in other sectors (Hancock & Walsh 2016), resulting in a situation whereby labour market transition is arguably a determinant of personal inclinations or occurring opportunities. Notoriously, in Italy one's social capital play a relevant role. Social capital, unfortunately, is seldom measured in this topic, and it probably revealed interesting patterns about PhD-holders' eventual employment trajectories. This study to this regard provides some proxies of social capital to disentangle why a fresh PhDholder might happen to stay in academia or to opt for another industry. These proxies derive from different types of contacts a PhD-holder may experience during and after PhD attainment.

Realising one's chances.
It is reasonable to argue that for PhD-holders any non-academic position is still today perceived as a sort of exit from an academic career. This may happen once a person has pursued research for research's sake and after having accomplished this interest. More likely, to continue to carry on one's research sooner or later may clash with the issue of what to do for a living. In this scenario, PhD-holders realise at different stages that only few people will have the opportunity to do research forever, and, more importantly, PhD-holders realise at different stages whether they are among those few achieving a career in academia. We also cannot assume that any PhD-holder is able to realise his/her actual chances promptly and with no biases. They also may realise only after a given period of time the extent to which they really want to stay in academia, or under which conditions they wish to continue to be academics (Sauermann & Roach 2012). Only occasionally, research casts light on preferences for salary and career as such, at disfavour of research for research sake (Roach & Sauermann 2010). This problem is exacerbated by the fact that a PhD-holder do not accrue particular advantages in the labour market outside academia in comparison to a Master graduate, at least for the private sector (Pedersen 2016). The conundrum of having spent years in doing research and finding oneself in a precarious and/or dead-end career trajectory lasts.
As a result, the literature is still not fully developed in analysing the employment opportunities for PhD-holders in the mid/long-term, leaving space for essential research to understand, for instance, which skills ought to be developed during PhD studentship for the sake of future employability, or to understand which events at the aftermath of doctoral attainment are more likely to help PhD-holders finding a good job in non-academic sectors.

Data
The analysis is based on a secondary dataseta dedicated survey about PhD-holders' employment destination from any discipline released from Italian statistical office (Istat, Indagine sui dottori 3 ). The data refer to PhD-holders who obtained their qualification in 2008 and 2010 in any Italian institution. This study cannot analyse career paths across a long time span and with events in a timeline like Webber & González Canché have done (2018). Nevertheless, the data afford to identify some events such doctoral studentship, post-doctoral positions (either on-going or accomplished) and current employment, along with scientific productivity, participation in projects, and mobility. Descriptive statistics of the dataset are presented in Table 1.

Dependent variables.
Economic sector of employment is the first dependent variable. To minimize this variable into a few coherent categories, we disentangle two possible different sectors other than the academic-scientific one: other education, and all the rest. This distinction is relevant because education can be considered in Italy a second-best-alternative for PhD-holders who might not have found a stable employment in academia after a certain number of years. Some PhDholders thus may opt to teach, using their knowledge and qualification for this career. In comparison to academic employment, these jobs are on average lower in wages. All the other sectors instead include anything, and they are more likely employment destinations with more diversified demanded skills. In principle, PhD-holders should spread in many different sectors regardless of the R&D intensity of a sector. PhD-holders in other sectors, in fact,either public, private or "third sector" (i.e. quasi-governmental, quangos, associations, etc.)may arguably use their skills and knowledge from any discipline in more innovative ways, filling R&D positions in any sector. For this reason, any employment in academia is compared against these two possible destination sectors separately, using the same set of independent variables in hypothesis 1 (horizontal segmentation, see below). The analyses exclude PhD-holders who already were in employment when they started the PhD in order to avoid the effect of possible improvement of qualification on the evidence that these PhDholders usually do not change sector, nor even employer, when the PhD is obtained (figures available upon request). The second dependent variable used for hypothesis 2 (vertical segmentation, see below) is salary [WPPP; average 64.46€ net per day] (See Table 1). Considering the net monthly income declared and the respective weekly worked hours, it is possible to compute this variable. In particular, this variable has been generated taking into account the net income of different possible taxation regimes, and after having compared the income at purchasing power parity (PPP) of the country of residence, applying coefficients available from OECD figures for the year 2014. Yet, this variable also considers further possible income from any other work, which is relevant for several PhD-holders active in flexible working conditions. This portion of the income is considered proportionally to the time spent on average in order to have a fair overall average of wage per unit of time. Some supplementary benefits are included, such as tredicesima and quattordicesimatypical monetary benefits for employees in Italy. Since the variable is about salary over working time, we overlook whether parttimers are not full-timers on a personal preference or not (around two thirds of respondents declared they would have preferred not to be part-timers). We reckon this caveat a minor limitation. The test for the second hypothesis regresses the natural logarithm of this variable.

Independent variables.
Independent variables might be grouped in some groups such as biographical information, doctoral experience, employment after PhD attainment (either post-doctoral or other), and current employment. Some of these variables are discussed to facilitate the results interpretation. Funding of PhD stipend [d1_10] explains if the PhD was covered by any stipend, and, if yes, which type of source the stipend was based on. This variable is interesting to understand if at this studentship stage contacts with non-academic source of funding may predict a future career in non-academic positions (Horta et al.2018 (Table 2 and Table 3), these latter two sets of variables have been normalized by number of years elapsed after PhD attainmentthe dataset incorporates two close-range cohorts, those who got the PhD in 2008 and those who got it in 2010. The variable about participation in projects, like that of source of funding of PhD stipend, is relevant to understand whether some types of organizations are more likely to serve as a gateway to let a PhD-holder get a job in sectors other than the academic or educational ones. These variables referring to participation in projects are categorical: no projects, national projects, international projects. They are kept as continuous on the assumption that international projects are more important, more prestigious, and may convey more social capital than the national ones. Completion of post-doc [expostdoc; 20.5% of sample] is a binary variable about any possible completed experience as a post-doc. This variable includes the typical Italian post-doc position (assegno di ricerca), such as other atypical contracts usually occurring to fund relatively short fixed-term contracts in academia. All these contractual forms are not necessarily propaedeutic to achieve any permanent position as assistant professor, or even the more recently established in the early 2000s fixed-term assistant professorship (ricercatori a tempo determinato tipo A and B). Possible channels through which PhD-holders got the current job are also available [d2_36 series]. The base in Table1 is set to "formal competition" because this is the way academic positions are obtained in Italy, whereas for other sectors (especially the non-public one) any other option more likely applies. Table2 and Table3 foster a simplified version of channels to get jobs. Variable "PhD title in getting job" [d2.39; 43% essential; 37% desirable and 20% not useful] and variable "PhD actually useful in job" [d2.40; 46.5% yes] further describe the relationship between PhD and current occupation considering the moment of interview. These variables can be relevant in understanding the capacity by employers to realise the actual value of PhDholders. A combination of these two variables may individuate a possible bias in not including PhD title at least as a desirable criterion when selecting staff. Table 1 around here

Hypotheses and results
The dataset as introduced is suitable to test two hypotheses.
Hp1. Specific doctoral and post-doctoral experiences determine the extent to which PhDholders remain in an academic career, with no specific role of teaching experience or research performances. Hp1a. These predictors are different between PhD-holders who exit academic career toward other education employment and those who work in other economic sectors. Hp2. PhD-holders have significantly different wages per unit of time according to specific doctoral and post-doctoral experiences, with no specific role of teaching experience or research performances.
The first hypothesis about sector of employment is tested with two separate multilevel mixedeffects logistic regressionslayers defined by all main 14 disciplinary categories used in Italian higher education system. This hypothesis helps individuate key factors in facilitating horizontal segmentation, assuming that it is useful to understand what does facilitate PhD-holders finding alternative careers to academiaeither fall-back in non-academic education (Table 2, Model1), or non-education at all (Table 2, Model2). The second hypothesis predicts income expressed as net income per unit of time. At parity of sector of employment and of academic discipline, it is relevant to understand what may predict a more successful use of one's PhD title. This second test can be referred as vertical segmentation, on the ground that typical and traditional academic careers use to have uniformed pay scales. This hypothesis accounts for two layers of multilevel analysis: the three destination sectors used for hypothesis 1; disciplines as per Hp1. This choice is justified by the evidence that labour market of PhD-holders by salary varies considerably by type of sector also within a single discipline (McFall et al. 2015). For all regressions, tests were launched accounting for heteroscedasticity in residual distribution (robust standard errors). Also possible multicollinearity is tested, resulting in having no problems (VIF scores for each variable below 2.35, average 1.41).

Hp1. How does leave academia?
The results shown in Table 2 are split between what does predict a move from academic employment into other education on one hand (Model1), and, alternatively, what predicts a move from academic employment into any other economic sector (Model2). This distinction is justified by the average salaries, but also by the assumption that pursuing other noneducation sectors as an employment destination implies more gumption in comparison to seeking employment in the 'other education' sectoran argument in line with Baruffaldi et al. (2017). It is in fact expected that at least to some extent "to teach in schools if one couldn't make what he/she studied a real academic career" is a "lazy" option. The results give substantially different explanations in comparing these two sector destinations. Observing Model1, PhD-holders are more likely to get a job in the 'other education' sector if their performances as undergraduates were successful [d1_7]. Nevertheless, there are no other predictors of this outcome related to biography or descriptions of doctoral experience. In terms of scientific productivity, there are no statistically significant coefficients. Participation in projects funded by universities [d2_47 series] is instead a strong predictor for not abandoning academic positions, possibly also the precarious ones. As expected, PhD-holders whose destination sector is "other education" are substantially less likely to have moved abroad [d2_24]. Overall, the profile of the PhD-holder who dropped from academia into other education seems to be that of an excellent student who did not find the way to participate after PhD-attainment to academic life and moved into a non-research position, possibly also because of preferring to stay nearby one's residency. Model2 gives different results. People who went to work outside the academic career or other educational sectors are more likely to be older (possibly because people try to persist in academic positions), males, and with more educated mothers. They are also more likely to have spent more time to complete the doctorate [d1_15], but they were already working at the moment of completion [d2_52]. Table 2 around here Even more so than is the case in Model1, these PhD-holders are less likely to have continued an academic career if they joined any project funded by universities. They are also more likely to have worked in academia under a project funded by private organizations [d2_47e and d2_27f]. There is also a very strong prevalence in becoming self-employed [10.d2_36]. In comparison to Model1, those who move on non-education sector are less likely to pursue research, but this is sensibly less prominent as a predictor. Overall, the typical PhD-holder described in Model2 is a person who is more likely to have worked with private stakeholders when still in his/her years spent in university. Moreover, they are more likely to become selfemployed. They also did not always have to give up research entirely [d2_24]. Table 3 shows the results of the second hypothesis about better jobs in terms of salary. Multilevel regression is apt in accounting for the assumption that wages are a function also of sector, and not only of discipline of doctoral study. Notably, the types of projects held after PhD attainment are relevant in predicting income more than other factors such as publications: projects funded by universities predict lower income; projects funded by private firms or international organizations predict higher income. Post-doctoral mobility is also associated with a higher salary [mob_post]. Direct call from employers, andabove allself-employment as channels to access PhD-holders' current jobs are the strongest predictors for having higher income. This fact is relevant as self-employment can be either very munificent or precarious, especially in the Italian context. This finding is useful if non-linear effect between entrepreneurship of skilled workers and respective push effect enacted to avoid unemployment is taken into account . This finding compels to investigate in future studies whether self-employment is incentivised by policies or it is a purely autonomous endeavour.  (Horta et al. 2018): funding of PhD studentship (stage0); funding at postdoctoral period (stage1); current employment (self-employment mode; stage2). Annex reports these tests by hypotheses, endogenous variable, and instrumental variables. Types of projects funded and held with private and international organization (stage1) determine selfemployment (tests #6, 11, and 12), causing a PhD-holder to be more likely to become a successful self-employed PhD-holder, especially when PhD-holders do not work in other education sectors. These tests demonstrate a link between working as a PhD-holders in partnership with companies and other organizations, and a consequential increasing likelihood in establishing one's own business, which is coherent with literature (Folta et al. 2010). Funding of PhD (no scholarship during PhD studentship -stage0) is determinant of exiting academic career into other sectors (test #8), but not other educationreinforcing the idea that horizontal segmentation is a bifurcation between other education from one side, and other sectors on the other side.

Discussion
Several aspects upon results deserve discussion against empirical literature. The most interesting finding pertains to funding players and the types of collaborations universities may have with companies or other organizations. PhD-holders are definitely more likely to get a job in other sectors than (broadly speaking) academic ones when PhD-holders are involved in projects funded by enterprises or international organizations at post-doctoral stage. Notably, this is not true when these types of potential future employers fund doctoral stipends, which is coherent with findings about funding PhD studentships (Horta et al. 2018). This is relevant also in relation to the highest coefficient foundthat of becoming selfemployed predicting higher salaries (vertical segmentationhypothesis 2). This mode of employment is found to have endogeneity with specific players funding projects at post-doctoral stage, such as companies or other international organizationssuggesting a specific Italian labour market pattern. Not necessarily other European labour markets are similar in relation to self-employment and how and why one person chooses to establish one's own business or professional activity. Arguably, the relationship with non-academic organizations could favour the absorption of PhD-holders by means of recruiting them in a second instance, if another context with different labour market pattern were taken as a case. The paper thus offers novel evidence to support arguments in the literature about the importance of partnerships with other sectors (non-higher education and non-public research institute) for increasing PhD-holders' chances to develop one's career. For some funding players, in fact, collaborations in the form of funding projects is a kind of investment. In this light, this study confirms Garcia-Quevedo's (2012) findings in terms of likelihood of hiring PhD-holders, but it adds that the best matching occurs at the end of post-doctoral relationships. A fair interpretation is that projects with post-doc holders probably allow PhD-holders and nonacademic employers to make sounder reciprocal acquaintance. Arguably, they also let a given PhD-holder understand how to convert his/her scientific knowledge into a viable business, which does not necessarily exclude research in toto. In this light, our analysis resonates the idea that self-employment is more likely to be a smooth transition from wages into entrepreneurship (Folta et al. 2010), rather than a sharp decision. PhD-holders find themselves increasingly within specific, fixed-term, projects (Nielsen & Cappelen 2014;Cantwell & Taylor 2015;Signoret et al. 2019;Hokanson & Goldberg 2018;Teelken & van der Weijden 2018) anyway. Ultimately, these results glimpse viable solutions to this well-known phenomenon, suggesting that projects with a link to external world are valuable opportunities to develop a better next step for one's career. In terms of sector of employment (horizontal segmentation), the paper finds that men are more likely to get a job in other sectors, which is in line with other findings on the interplay between gender, tenure and getting jobs (Rudd et al. 2007). Coherently with other studies (Webber & González Canché 2018;Evers & Sieverding 2015), there are no discernible differences in gender when it comes to define employment destination of PhD-holders. The vertical segmentation (differences in salary) is as well relevant. Whilst post-doc periods don't appear to have a pay-off (Recotillet 2007), women are more likely to suffer the family burden, although there is no gender pay gap at the highest levels of income. This study also finds no relevance of parents' education in predicting higher income of PhD-holders, as was the case in findings from a study in Spain (Caparros-Ruiz 2019) and Chile (Chiappa & Perez Mejias 2019). Mobility does play a particular role. The findings corroborate the results by Caparros-Ruiz (2019) and Bonnard et al. (2017) in interpreting scientific mobility as a behaviour apt to increase scientific opportunities. It is possible that academic positions nowadays have to be so specific that the probabilities to find them elsewhere from one's university of PhD attainment are high. This pattern tends to spread academic job opportunities across an international ray. Mobility, in other terms, could mean a search for better conditions (Caparros-Ruiz 2019; Di Cintio & Grassi 2016), and it does not necessarily generate beneficial effects in terms of wage. This study also finds that mobility does not predict sector of employment, which tends to reinforce the aforementioned interpretation. The overall interpretation regarding mobility is similar to that of recent debate about Italian PhD-holders and the respective brain drain (Passaretta et al. 2019;Cattaneo et al. 2019). However, mobility after PhD attainment positively predicts higher salaries at parity of cost of life in country destination, which does not happen with mobility during PhD. This difference could be read analogously with findings about funding players.

Conclusion
Nowadays any practice favouring an exit from academic labour market for PhD-holders is not just an option. It is an increasing necessity. The destiny of PhD-holders is that of getting jobs like any other graduate, otherwise biases about the poor contribution of PhD-holders to the whole economy (Santos et al.2016) might continue to recur. To this respect, the paper contributes not only in confirming that PhD-holders can be useful for society and for PhDholders themselves, but it gives novel insights about: a) how getting a job in the nonacademic labour market is more likely to happen; and b) when PhD-holders more likely yield the most from their highly skilled profiles. From this analysis, segmentation of PhD-holders' employment destinations can be predicted both at studentship and especially post-doctoral stages due to specific non-academic projects funders. Attempting a more general understanding of the topic, PhD-holders are more likely to benefit from situations where awareness about science-in-the-making (e.g. research funded by private organizations or other organizations) and respective impact can stimulate other career pathways, as a seminal study already highlighted (Mangematin 2000). For the particular Italian context, this happens via entrepreneurial behaviour, but for a more tentative generalized conclusion at European level one may think about incorporating novel engaging practices to favour both supply and demand sides of labour market. These engaging practices are likely to realise reciprocal benefits. Many funding schemes at national and European regional level might refer in the future to possible conversions of research activities into innovations as part of evaluation criteria. Such evaluations may favour the matching between PhD-holders and projects stakeholders. Self-employment is a valuable option, although this is only one possible modes of conversion of PhD-holders' highly specific knowledge and expertise into viable nonacademic employability. The practice of considering non-academic funders as drivers of segmentation could bring about reciprocal benefits to all: universities that might propel more projects; employers who might discover unexpected human resources from "young talents"; PhD-holders themselves who increasingly will have to demonstrate what a PhD in their portfolio may bring for the "real world".  Scientific production: patents -0.3820 *** -0.5544 *** d2_47a Funding Mobility after PhD -0.1882 -0.0674 Self-empl.