1 Introduction

Although in “a time of unparalleled technological development, it is the human resources that paradoxically spell success or failure for all firms, and especially entrepreneurial ones” (Katz et al., 2000, p. 7), the literature is just beginning to examine employee retention in young ventures and what determines turnover intent. This is an important research topic because “few imperatives are more vital to the success of young technology companies than retaining key technical personnel” (Baron & Hannan, 2002, p. 21). Initial studies find that promotion possibilities are associated with higher retention rates in small ventures (Heneman & Berkley, 1999; Kemelgor & Meek, 2008) and that ethical leadership (Theriou et al., 2020) and job embeddedness (Coetzer et al., 2019) are negatively associated with turnover intentions. Gialusi and Coetzer (2013) find that reasons for intended and actual turnover among employees in small firms are relationship conflict and lack of career prospects. Moreover, the characteristics of the entrepreneur (e.g., the entrepreneur’s displays of passion, Breugst et al., 2012) can influence employees’ commitment to young ventures.

However, despite these initial insights, one limitation of prior work on employee retention is that it has insufficiently considered that over time, young ventures are likely subject to substantial adaptation and change. From an employee’s perspective, one key difference between working at a young venture versus working at an established firm is that while the latter have rather stable structures, identities, and other attributes that attract employees in the first place, these attributes are highly dynamic in the first years of a young venture (Fisher et al., 2016; Greiner, 1972). For example, as the venture matures, initially flat structures become more hierarchical, and informal communication channels and cultures become more formalized (Greiner, 1972). Furthermore, maturing ventures often adjust their organizational identity to accommodate changing stakeholder needs (Domurath et al., 2020; Fisher et al., 2016). Hence, young venture employees likely actively try to make sense of venture changes (cf. Louis, 1980) to determine if those changes meet their expectation of satisfying at least their basic need of continued employment and financial security. Thus, while an employee may initially find working in a young venture attractive based on the venture’s attributes at the time they apply for a job, turnover intent may increase over time due to the changes that the venture experiences.

Although it may not always be desirable for a venture to keep a specific employee, understanding what triggers employees’ turnover intent in young ventures is important because (1) turnover intent has been positively linked to employees leaving the firm (Allen et al., 2005) which can be associated with the loss of critical human capital, (2) young ventures have only limited resources for recruiting new personnel (e.g., Hornsby & Kuratko, 1990) to replace those who have left, (3) employees who have turnover intent, but do not quit, represent a crucial waste of resources for the venture because these employees show low job satisfaction (Hom et al., 2017; Rubenstein et al., 2018) which in turn is associated with high absenteeism (e.g., Brooke & Price, 1989), and (4) these problems are particularly prevalent given the generally growing workforce mobility (e.g., Raffiee & Byun, 2020) which taxes the ventures’ sparse resources. Therefore, we ask the following: To what extent does employees’ turnover intent in young ventures change over time, contingent on employee prior start-up work experience and venture growth?

To address this question, we draw on prior research on turnover intent. While this research has examined a long list of variables correlated with an individual’s turnover, “extant models explain no more than 20% of turnover variance” (Klag et al., 2015, p. 37). Hence, attention has turned to theories focused on the thought processes related to turnover intent. Therefore, to explore turnover intent, we draw upon one such theory—met expectations theory. It proposes that the discrepancy between what an individual expected to encounter when entering a venture and what they actually encounter influences turnover intent (Porter & Steers, 1973). Thus, met expectations theory is particularly suited for the purpose of our study because it accommodates the potential changes young ventures experience over time and the effect of these changes on turnover intent. Empirically, we draw on a unique, hand-collected, and longitudinal data set based on 1,151 questionnaire responses from 458 employees of 67 ventures over a 1-year time frame. The findings of our study offer novel implications for the literatures on young venture employees, venture growth, and turnover intent.

First, we contribute to the discussion on employee retention in young ventures (Cardon & Stevens, 2004). Research on established firms emphasizes that employees adjust their expectations to decrease turnover intent over time (e.g., Major et al., 1995). Adjustment processes are facilitated by HR practices that help employees manage expectations (e.g., orientation programs, accurate job analyses and job descriptions, clear career paths), reduce “reality shock,” improve an employee’s ability to cope with the new job demands, and facilitate aligning of personal job values with job experiences (e.g., Baur et al., 2014; Ślebarska et al., 2019; Wang et al., 2017). Thus, employees update and adjust their expectations to the stable structure, norms, values, and culture of their organization reducing turnover (e.g., Bauer et al., 2007; Major et al., 1995; Peterson, 2004). By showing that young venture employees’ turnover intent increases (rather than decreases) over time, our work indicates that the theoretical mechanisms explaining turnover intent in large organizations may not apply to young ventures due to the ventures’ constant evolution and frequent change along with immature HR practices and evolving organizational identities. Therefore, our study emphasizes the theoretical distinctiveness of young ventures in comparison to established firms when studying employee turnover intent.

Second, we identify conditions under which employee turnover in young ventures (and the theoretical distinctiveness between young and established organizations) may be particularly severe. While prior research identifies the importance of employee retention for venture growth and survival (Gjerløv-Juel & Guenther, 2019), our research shows that young venture growth can also be a prerequisite for retention. Moreover, while prior work has identified experience with founding and working in a young venture as a critical determinant of entrepreneurial firm outcomes (e.g., Gilbert et al., 2006), including venture survival (Xi et al., 2020), we demonstrate that employee start-up work experience impacts their turnover intent over time and thus the young venture’s human capital endowment (contingent on the venture’s growth rate). Therefore, we shift the attention away from founders’ experience to suggest that work experience in entrepreneurial firms plays a critical role for employees’ attitudes toward their work, including turnover intentions. In sum, we suggest that contingencies at both the level of the individual employee and the organization (conjointly) impact to what extent theorizing on employee turnover must take potential differences between the young venture and established firm contexts into account.

2 Theoretical development

2.1 Met expectations theory

Human resources (HR) management practices and human capital are key drivers of firm growth (Demir et al., 2017). Most previous HR research in young firms focuses on recruitment, selection, and onboarding practices rather than on employee retention (Cardon & Stevens, 2004; Demir et al., 2017), yet employee retention is critically important to young ventures because employees’ “knowledge often represents the firm’s most valuable asset” (Baron & Hannan, 2002, p. 21). Hence, we focus on what influences turnover intent in young ventures. Turnover intent is an employee’s desire to voluntarily separate from an organization (e.g., Hom et al., 2017). As a criterion, it has advantages over behavioral measures because actually leaving a venture is often influenced by numerous other exogenous variables (for example, ability of a spouse to relocate, better job alternatives, economic conditions, e.g., Allen et al., 2005; Hom et al., 2012; Mobley et al., 1979) that often cannot be controlled by the investigator, but influence research findings. Turnover intent is a continuous variable (rather than a dichotomous variable like actual turnover) that reflects an employee’s attitude toward the organization. Nonetheless, turnover intent leads to job searches, and when an alternative job presents itself, turnover often occurs (Hom et al., 2012; Mobley et al., 1979). Meta-analyses have shown significant correlations between turnover intent and actual turnover (Griffeth et al., 2000, ρ = 0.38; Wong & Cheng, 2020, ρ = 0.29). When turnover intent is declared, one can assume some action will be taken by the employee. “Given its predictive superiority […] turnover intentions have served as a surrogate or proxy for turnover when quit data are unavailable” (Hom et al., 2017, p. 533).

In explaining turnover intent, met expectations theory (Porter & Steers, 1973) suggests that employees make comparisons between their pre-entry expectations (e.g., expectations developed during recruitment, selection, and socialization processes) and post-entry experiences. When post-entry experiences fall short of pre-entry expectations, turnover intent increases. The theory addresses a common practitioner concern about how activities at early stages of the employee life cycle (attraction, recruitment, and onboarding) influence retention. The more an individual’s actual experience in the venture falls short of their initial expectations at a given point in time, the lower their intrinsic motivation (Gkorezis & Kastritsi, 2017), job satisfaction (Maden et al., 2016; Wanous et al., 1992), organizational commitment (Wanous et al., 1992), and the higher their turnover intent (Maden et al., 2016; Wanous et al., 1992) and turnover (Earnest et al., 2011). That is, individuals search for, and take, employment that sufficiently meets their needs, values, and preferences and when they find those needs, values, and preferences are unmet, they report greater turnover intent. Moreover, Turnley and Feldman (2000) report that the relationship between employees’ beliefs regarding the terms and conditions of an exchange agreement they had with their organizations and their intent to quit was partially mediated by unmet expectations. One main reason for unmet expectations is unrealistic pre-entry expectations (Buckley et al., 1998) which, as we will suggest below, may impact employees’ turnover intent over the time they work in young ventures.

2.2 Employees’ turnover intent in young ventures over time

To date, few studies have examined how key characteristics of a young venture may affect turnover intent. Past research shows that employees are concerned about HR-related inducements and expectations (e.g., job security, opportunities for advancement and compensation, Batt, 2002; Turban, 2001). However, young ventures may be more likely to be perceived as not honouring the promises because they have underdeveloped HR departments and policies (e.g., Hornsby & Kuratko, 1990; Massey & Campbell, 2013; Williamson, 2000). That is, they have fewer resources and incentives than large firms to identify and fulfill the inducements and expectations perceived by new hires during the recruitment and selection process. Indeed, perceived unfilled promises have been associated with higher turnover intent among employees in small- and medium-sized businesses (Kickul, 2001). Also, when compared to established organizations, young ventures are characterized by constantly evolving unwritten practices, routines, and tacit understandings and influenced by social relations and the prominence of informality (Wapshott & Mallett, 2013). Presumably, to some extent, young venture hires are excited by and value such a workplace environment, yet young ventures are also characterized by dynamism and change, which can lead to a workplace environment that employees find less exciting and attractive.

As a venture matures, it changes “in a series of configurations” that reflect dynamic states of a match between business model and environment (Levie & Lichtenstein, 2010, p. 335; see also Phelps et al., 2007). For example, Greiner (1972) describes how young ventures over time have to adapt their organizational practices to accommodate the new internal environment and enable further development. For example, during the very early phase of venture development, founders tend to have a strong technical focus, are primarily concerned with creating and selling a new product or service, and communicate with employees frequently and informally. However, as the venture matures, “employees cannot be managed exclusively through informal communication” (Greiner, 1972, p. 42). Therefore, management practices must change. Similarly, Kazanjian (1988, p. 263) reports that ventures in the conception and development stage focus primarily on inventing a product or technology with structure and formal processes being “nonexistent during this stage, with almost all activity focused on technical issues as defined and directed by the founding entrepreneur(s).” However, later organizational functions and administrative systems emerge with more hierarchy and formalization of positions (Kazanjian, 1988) and the differentiation of the labor force increases (Masurel & van Montfort, 2006). Indeed, the changing types and patterns of specialized functions emerging in young ventures reflect the most prominent management challenges (Hanks & Chandler, 1994).

Finally, Fisher et al. (2016) argued that as young ventures mature, they need to appeal to different stakeholders, which have different norms, standards, and values (see also Jawahar & Mclaughlin, 2001). Thus, young ventures must adapt their organizational identity to be perceived as legitimate. For example, during the early conception stage of a technology venture, technological inventions are frequently based on academic research. Resource providers are grant administrators and scientific experts who primarily evaluate the potential for technological breakthroughs and societal benefits rather than commercial outcomes. As commercial outcomes are very uncertain, ventures will rely on the status of their institution and team members to enhance legitimacy. However, when commercialization becomes more important, venture capitalists and angel investors are potential sources of funding. These investors will look for promising investments based on financial, competitive, and team data. Therefore, young ventures have to adapt their identity over time to appeal to their varying stakeholders and this can lead to feelings of grief and tragedy tied to personal identity loss in employees (cf. Hay et al., 2021).

In sum, literature suggests that when compared to established organizations, young ventures (a) have underdeveloped HR departments and fewer HR resources, (b) experience more organizational change, and (c) are characterized by evolving organizational identities. These characteristics of young firms may negatively impact employee turnover intent. Gjerløv-Juel and Guenther (2019, p. 84) suggest that although numerous explanations exist for employee turnover in the initial years of a young venture (e.g., employee burnout, conflict, and new opportunities for experienced employees like founding their own firm and being recruited to other start-ups), one reason for turnover intent is employees initially “attracted by the innovative and vibrant growth environment” becoming despondent at the development of procedures, routines, and hierarchies as a firm increases in size. In the face of the uncertainty and complexity inherent in the start-up experience and a lack of HR practices to narrow the gaps in expectations between a new employee and the firm, employees likely try to gain a sense of control by making sense of what the changing work environment means to them in terms of meeting their expectations. As a venture matures and changes, perceived unmet expectations likely arise in employees. For example, when an employee decides to work for a venture because the typically quite communal team climate (e.g., Tumasjan et al., 2011) fits with their personal needs and expectations, a change from an informal management style toward a more formal, impersonal management style that uses a hierarchy of titles and positions (e.g., Greiner, 1972) leads to the loss of an organizational factor that made organizational entry attractive in the first place. Feelings of unmet expectations may result and this is reflected in increased turnover intentions.

Similarly, as a venture progresses from conception to commercialization, employees who rate a venture attractive because of its scientific focus and reputation might not experience congruence with the identity of a venture that seeks to increase sales and attract venture capital. Thus, even if the employee still performs their original job task within the venture (e.g., as an engineer), there may be unmet expectations, and an increased turnover intent for the employee. Therefore, we propose the following baseline hypothesis in recognition that constant evolution and frequent organizational change, along with immature HR practices managing expectations, are often synonymous with the start-up context and may increase the chances of unmet employee expectations:

  • Hypothesis 1: In young ventures, employees’ turnover intent increases over time.

2.3 Employees’ turnover intent and prior start-up work experience

Research suggests that the extent to which unmet expectations form is influenced by whether initial expectations were realistic (Buckley et al., 1998). Employees with realistic expectations are more likely to have those expectations met. Previous experiences in a similar job serve as an important determinant of future job expectations (Carr et al., 2006; Rousseau, 2001). For example, individuals who have experienced organizational change in previous work relationships (including downsizing, restructuring, and bankruptcy) show lower expectations of security in their current job (Cavanaugh & Noe, 1999). Hence, as employees’ current job expectations are influenced by prior job experiences, they are more likely to form realistic expectations when they have previously gained experience in an organization that is similar to what they are currently experiencing in a young venture.

We therefore expect that employees are better able to form realistic expectations for the current job in a young venture when they have accumulated experiences and knowledge about how young ventures change over time. Specifically, since experience builds with time, we expect that employees’ job expectations in their current venture are more realistic the more time they have worked in other start-ups previously. The longer employees have worked in other start-ups prior to the current one, the more they are familiar with the fact that change is part of the development of any young venture. These employees will therefore have the expectation that change will be part of the current job as well. Hence, these employees are more likely to accept and embrace change experienced as their current venture develops over time, rather than being surprised, scared, or dissatisfied by change. Thus, at least in the short- and mid-term, employees with more extensive work experience in prior start-ups are less likely to face a situation where their expectations related to change within their current venture environment are unmet.

Moreover, a young venture may not be able to provide realistic job previews based on the uncertain nature of young venture development, which makes it more difficult for employees to develop realistic expectations. For example, Mayson and Barrett (2006) report that only 53% of the small ventures in their study provided detailed previews in a written manner for positions they wanted to fill. In sum, we expect that when entering a young venture work environment, the more time employees have spent in other start-ups prior to their current one, the more they expect, and the better they are prepared for, the changes and challenges they are likely to face over time in their current venture. That is, we suspect that the number of unmet expectations, and resultant turnover intent, diminishes with the amount of prior start-up experience. A hire with previous comprehensive work experience comprising all maturation stages of a young venture may have realistic expectations over time and no intent to leave. At the other extreme, a hire without any prior experience of venture maturation may perceive unmet expectations with every perceived organizational change. However, most hires will likely fall between the two extremes. As a result, we expect that employees’ turnover intent increases less over time for those employees who have longer prior start-up work experience. That is, we propose that the more work experience the less the turnover intent increases over time. Therefore, we propose the following interaction hypothesis:

  • Hypothesis 2: In young ventures, the increase in employees’ turnover intent over time is weaker for employees with longer work experience in start-ups prior to the current one.

2.4 Employees’ turnover intent and venture employee growth

Maden et al. (2016) suggest that employees’ future expectations affect their responses to presently unmet expectations because their projections of future work conditions shape their interpretation of current unmet expectations and influence their self-motivation. Proost et al. (2012) find that despite their currently unmet expectations, employees who perceive learning opportunities in their current job, including opportunities for personal growth and development, are less likely to leave their job. In addition, when employees perceive a discrepancy between expectations and employer actions, employees’ beliefs in “things getting better” in the future (Bankins, 2015, p. 1087) can act as a remedy to the employment relationship.

Therefore, besides personal factors, organizational factors likely moderate how turnover intent develops over time and some of these factors may be perceived as positive so as to mitigate turnover intent. For the unique young venture context, growth plays a key role and may be potentially perceived as a positive change factor. Given the high failure rates of young ventures (Brüderl et al., 1992), employees working at young ventures likely look for signs of venture success to alleviate their concern that the venture may fail. Venture growth suggests venture survival due to reduced average costs, reaching minimally efficient scale, access to capital, the ability to attract talented employees, legitimacy, and benefits from learning effects and efficiency gains (Gjerløv-Juel & Guenther, 2019). Hence, it is not surprising that achieving high growth is one of the most important goals for many entrepreneurs (Wiklund & Shepherd, 2003). Consistent with the large body of literature on young venture growth, we expect that growth is a particularly important indicator of “positive change” in the young venture context. Indeed, Rutherford et al. (2003) found that problems in managing a venture’s human resources vary with venture growth such that high-growth ventures have the least problems with retention and compensation whereas no-growth ventures face the most severe recruiting problems.

Scholars conceptualize and measure growth using multiple dimensions including the increase in revenues, profits, assets, equity value, and staff complement (Shepherd & Wiklund, 2009). In particular, young ventures in high technology industries often do not generate substantial (or any) revenues and profits, and they are not listed on the stock market such that their valuation is difficult and often non-transparent. Therefore, the change in the number of firm employees is a critical indicator of performance in these ventures (e.g., Hanks et al., 1993). Also, due to the small size of young ventures, employee growth is typically visible to all employees and therefore can serve as a tangible sign of venture performance to employees, while a decreasing number of employees is likely seen as a sign of crisis and potential failure leading to higher perceived job insecurity among remaining employees, as well as negative performance implications and reputational damage for the firm (see Datta et al., 2010 for a review on effects of employee downsizing). This contrasts with financial indicators, which founders often keep secret and may only share with their investors, but usually not with employees. Indeed, Gjerløv-Juel and Guenther (2019, p. 89) in their study of young ventures found “a general positive survival effect of employment growth.”

Thus, we expect that periods of staff complement growth in a young venture will indicate to employees that the changes they experience represent opportunities and longer-term prospects (e.g., Bennett & Levinthal, 2017) that allow them to maintain (or even increase) need fulfillment. When ventures experience a period of employee growth, employees may expect need fulfillment in terms of secure employment, professional development, and increased responsibility. This may also result in employees feeling like they contribute to venture success because they possess characteristics that complement the organization and make them feel like they are unique and of value to the venture. In contrast, employees of ventures that experience a period of less growth, or even employee layoffs, are likely to perceive these changes as limiting need fulfillment (e.g., future opportunities for their own personal and professional development) and counter to their personal goals and values. They may even see the changes as a threat to their own future in the venture.

Past empirical evidence supports our arguments. Maden et al. (2016) show that when job expectations were unmet, employees with more positive future expectations experienced less emotional exhaustion and job dissatisfaction, in comparison to those with less positive expectations. They argue that employees “with less positive future expectations would interpret the existing unmet job expectations more pessimistically and perceive them as a precursor for further adversity in their job” (Maden et al., 2016, p. 8). However, job conditions such as learning opportunities may facilitate employee coping with the unmet expectations (Proost et al., 2012) and make them feel motivated by the prospect that future career goals could be met. Therefore, we propose the following interaction hypothesis:

  • Hypothesis 3: In young ventures, the increase of employees’ turnover intent over time is weaker for ventures experiencing periods with higher employee growth.

2.5 Employees’ turnover intent, start-up work experience, and venture employee growth

Although a venture might be growing, its growth rate is unlikely to be constant over a long period of time (i.e., over several years). Specifically, young ventures typically go through multiple alternating phases of high and low (or even negative) growth (e.g., Hanks et al., 1993). Oftentimes, a young venture’s period of high growth is only temporary (e.g., the months following a financing round or the acquisition of the first major customer), but the current growth may slow down or even become negative (e.g., when the money acquired in the prior financing round is about to run out and a new round has yet to be closed, or when a major customer is lost). Alternatively, the current slow growth (or even need to lay off employees) may be only temporary and subject to change. For example, after a new financing round (series A funding may last from 6 to 18 months) is closed, a venture may be able to build up its workforce again by hiring new employees. Furthermore, the need to concentrate on the market introduction of the most promising product candidate may necessitate the release of employees working on the development of alternative product candidates in the short-term (which is often conceptualized as a timeframe of about 1 year, Shepherd & Wiklund, 2009), but after successful market introduction, the venture may have sufficient resources to hire new employees for future growth.

To the extent an employee is aware of the potentially temporary nature of a venture’s period of high or low growth, they are less likely to consider this transitory development when making assessments of unmet expectations. Prior start-up work experience is a personal factor that provides employees with knowledge about the often-temporary nature of current venture development and therefore facilitates their building of more realistic expectations about the sustainability of the venture’s current growth period. These experienced employees are less likely to extrapolate current growth into expectations of the more distant future when making met expectation judgements. For example, during their prior work in start-ups, an employee may have personally experienced various ups and downs that characterize the start-up context (e.g., Bruno et al., 1992; Bygrave, 1993), or may have observed these ups and downs in other start-ups (e.g., in an incubator or accelerator environment). From both personal experiences and the observation of others, the employee has learned about the potentially temporary nature of high and low growth periods for the venture. Such an employee is less likely to assess expectations based on the current and short-term venture growth. Alternatively, the employee may incorporate the volatility of a venture’s growth into their expectations at the time they are applying for a job in the venture. That is, the employees may find the ups and downs they have experienced during their time in prior start-ups somewhat attractive or at least acceptable for their job, motivating them to select into the volatile job environment of their current venture.

In contrast, employees with little start-up work experience are less likely to be knowledgeable about, and prepared for the temporary nature of the (high or low) growth their venture experiences in the current period, and they are therefore more likely to make an unmet expectation assessment and extrapolate it into the future. For example, such an employee may expect that the current high growth rate of their venture will be sustainable, and the changes currently experienced within the venture will offer them multiple opportunities for personal and professional development in the future. This employee will be less likely to conclude that their expectations are unmet in their venture. Similarly, an employee with little start-up work experience who works for a venture that currently experiences a period of very little growth, or even must lay off employees, may not expect that the situation may change again in the future. Such an employee is likely to expect that the venture’s low short-term growth rate may eventually lead to sustained underperformance or even venture failure, such that the internal changes experienced over time do not offer possibilities for the personal and professional development that would form the basis of perceiving met expectations. For this employee, the current venture growth rate is likely to have a strong impact on turnover intent. Therefore, we propose the following three-way interaction hypothesis:

  • Hypothesis 4: In young ventures, the increase of employees’ turnover intent over time is weaker in ventures experiencing periods of higher growth, but this interaction effect is stronger for employees with shorter work experience in start-ups prior to the current one.

3 Methodology

3.1 Sample and data collection

To test the proposed model, we used a multi-source longitudinal survey study which included both the founders of young ventures and their employees as participants. Founders and employees answered questionnaires every 3 months over a period of 1 year (4 rounds). We expected that a period of 3 months between questionnaires would be appropriate because within 3 months ventures often experience substantial internal change (Domurath et al., 2020) that might impact employees’ turnover intent. Furthermore, in focusing on a 1-year observation period, we are consistent with many studies conceptualizing periods of short-term firm (employee) growth (e.g., Shepherd & Wiklund, 2009).

We used multiple sources to identify potential participants. First, we generated a list of potential participating ventures from the websites of entrepreneurship, technology, and incubation centers as well as from business angels and online entrepreneurship portals in Germany. We added only ventures that were 5 years old or younger (foundation in 2008 or later) to our initial list, consistent with other studies’ definitions of young ventures (e.g., Amason et al., 2006). Overall, we identified 1,296 ventures. However, through the process of contacting these ventures, we removed 731 because they no longer existed in their original, independent form, were subsidiaries of larger firms, or because they had already failed and closed operations.

Of the remaining 565 ventures, we contacted the founders personally, by telephone, or by e-mail to explain that we were conducting a study on young venture employees. In total, 128 ventures agreed to participate in the first data collection round, representing a response rate of 22.8%. Overall, the data collection yielded 1,830 responses nested within 696 individuals in 98 ventures. Since our study focuses on venture employees, we excluded the responses from founders, which reduced the number of observations to 1,202. Furthermore, we excluded observations with missing data for the focal variables of this study (a total of 51 observations). Hence, 1,151 observations nested in 458 employees from 67 ventures are the basis for our analysis.

At the beginning of the study, the average venture was 2.24 years old (SD = 1.36 years), employed on average 13.44 (SD = 15.37) people, and grew on average by 4.89 employees (SD = 14.34) over the observation period. Individual venture growth rates were on average 43.83% (SD = 75.57), with the worst-performing venture losing 80% of its workforce, and the best-performing venture growing by 300%. Employees were on average 29.44 (SD = 6.73) years old and 42.14% were female. They had worked for their current venture for 1.00 (SD = 0.96) year on average.

3.2 Survey administration

We collected data in four rounds over the course of a year. As an initial step, we asked founders who agreed to participate in our study to provide a list of all the employees in the venture. Before each data collection round started, we asked the founders of all participating ventures to update the list (i.e., with information on who had left the venture and who had joined after the previous data collection round). Once we had created or updated the list of potential participants, founders and employees received a personalized e-mail invitation to the survey. After 2–3 weeks, we sent a first reminder to all potential participants who had not participated.

In total, 311 out of 559 respondents in the first round completed the last survey while 248 dropped out, i.e., did not participate anymore (individual attrition rate = 44%). At the firm level, 75% of the ventures continued to participate until the end of the study, resulting in a firm-level attrition rate of 25%. The non-response rate describes how many participants who answered the survey in one round and were invited in the next round, but did not answer in the following round. The non-response rate was 18% between rounds 1 and 2, 21% between rounds 2 and 3, and 11% between rounds 3 and 4.

3.3 Variables

We designed the survey to measure individual and organizational variables over time. The instrument contained different parts. The first part contained measures of turnover intent and constructs that we assumed to change over time. This part was sent to all participants in each data collection round. The second part included measures of firm-level variables (e.g., number of employees) and environmental characteristics (e.g., environmental dynamism). We assumed that those variables would change over time and therefore they were measured in each data collection round. However, we also assumed that only founders would be in a position to appropriately answer these questions. Therefore, we sent this part of the survey instrument only to the founders participating in our study. Finally, the last part included measures of individual-level characteristics that were constant over time, such as participants’ previous work experience. We asked participants to provide this information only once.

3.3.1 Turnover intent

Although employees had left the surveyed ventures over the course of our study, we cannot include actual turnover in our analysis because we do not have full information on who left the companies. In addition, for most of the employees for whom we have exit information, we do not have survey responses prior to their organizational exit. Nevertheless, critical implications can be derived from insights about employees who report turnover intent, but do not leave an employer (please see Discussion below). We assessed turnover intent with three items based on Cammann et al. (1983)This measure is from the Michigan Organizational Assessment Questionnaire, a well-validated instrument for measuring employee attitudes at work. Scales from this instrument, including turnover intent, have been used in various studies and diverse contexts such as IT workers (Samgnanakkan, 2010), nursing, (Cheng et al., 2016), employees in the hospitality industry (Chan & Ao, 2019), and sales people (Chelariu & Stump, 2011). The items include “I often think of leaving the organization”; “It is very possible that I will look for a new job next year”; and “If I may choose again, I will choose to work for the current organization” (reverse coded). The participants expressed their agreement with the statements on a 7-point Likert-type scale ranging from 1 “strongly disagree” to 7 “strongly agree.” We granted participants anonymity to mitigate potential social desirability bias. The Cronbach's alpha was 0.80, which is of similar magnitude to that reported in previous research (Bouckenooghe et al., 2013: 0.77; Chen et al., 1998: 0.78; Shaw, 1999: 0.83).

3.3.2 Time

Time is the independent variable in our study. Since we measured the criterion variable, turnover intent, in each of our four rounds of observation, time ranges from 1 to 4. To ensure that this timeframe is appropriate to capture substantive changes within the ventures, we asked each employee to report in an open questionnaire field any major events for the venture that had happened since the last survey round. Participants described 1,463 events in total (including multiple reports of the same event within the venture). While not all these events were associated with significant changes for the ventures, many of them were. Moreover, these changes were both positive and negative for the venture. For example, reported changes included the loss of a major customer; the successful and unsuccessful closing of a planned financing round; the exit of key employees and/or founders; delays, technical problems, and failure of product development projects; changes in the venture’s business model; loss of important collaboration partners; and restructuring due to problems with the venture’s liquidity. Overall, these comments and reports support our assumption that the employees were experiencing a large number of (major) changes in their current ventures over the observation period of our study.

3.3.3 Start-up work experience

We asked all employees to report the number of years they had previously worked in start-ups. We excluded the current venture from the measure. In line with best practice for multilevel modeling (e.g., Aguinis et al., 2013), we group-mean centered this variable.

3.3.4 Venture growth

We used ventures’ employee growth rate over our observation period to assess the extent to which the ventures in our sample experienced a period of high or low growth. In the survey, we asked only founders to provide the number of employees in each of the four data collection rounds spanning 1 year of observation. To include the temporal change of the number of employees in the statistical analysis, different quantification methods are available. For example, subtracting the time 1 score from the time 4 score would provide a simple change score, yet scholars have criticized the use of change scores due to the minimal information they provide on change and the correlation of change scores with initial status scores (Rogosa et al., 1982). Therefore, we followed others (Chen et al., 2011; Liu et al., 2012) and estimated firm idiosyncratic Bayes slope estimates. That is, for each venture, we estimated a separate multilevel model where we included time as the predictor variable and number of employees as the criterion variable. The resulting slope estimate calculated across the measurement times indicates employee growth per venture over our observation period. We included this Bayes slope estimate as the variable “venture growth.” More positive values indicate a greater increase in the number of employees and more negative values indicate a greater decrease in the number of employees over time. We grand-mean centered this variable.

3.3.5 Control variables

At the individual level, we controlled for age and gender as they have been shown to correlate with turnover intent (e.g., Flint et al., 2013; Joseph et al., 2007). We asked participants to indicate their age (in years) and gender (coded 1 = female and 0 = male). Moreover, we included industry experience (in years) of the employees as it provides employees with knowledge on typical development cycles of their venture’s industry and might therefore influence their turnover intent depending on industry performance. Furthermore, employees’ level of education might impact their turnover intent in a venture over time. Wei (2015) shows that general human capital is positively related to turnover intent, such that employees with more human capital are more likely to show higher turnover intent because they usually have more outside options. We used a measure reflecting the educational level of employees with 1 indicating that the employee has attended university and 0 otherwise. Finally, we included employees’ organizational tenure (in years) because employees go through socialization processes over time which can reduce employees’ turnover intent (Bauer et al., 2007). Including this variable is important because we do not observe only the first year of employees’ time with their ventures; rather, employees might have stayed with their ventures for different periods of time (and even since the venture’s foundation) before the start of our study. We group-mean centered all individual-level control variables.

At the firm level, we first controlled for firm age because older ventures may have more established structures and routines that influence the likelihood that employees will leave due to internal adaptations (Greiner, 1972). Second, we included the initial number of employees (in time 1) as larger ventures tend to experience lower relative growth rates (e.g., Hanks et al., 1993). Finally, we controlled for environmental dynamism because it is reflective of different industries. We measured environmental dynamism with the 5-item scale described by Green et al. (2008). Measurement items included, “actions of competitors are generally quite easy to predict” and “product demand is easy to forecast.” In the survey, only founders assessed environmental dynamism. Founders indicated their agreement with these statements on a 7-point Likert-type scale ranging from 1 = “strongly disagree” to 7 = “strongly agree.” The internal consistency of the scale was α = 0.66, which is similar to 0.72 reported by Green et al. (2008). In case two or more founders participated in the survey, we used the mean of their answers for analysis. We grand-mean centered all firm-level variables.

Regarding potential common method bias, we note the different nature of our variables and their measurement (time is an objective variable that was not provided by the respondents, experience is also an objective value, and venture growth was provided by founders not by employees). This should minimize common method variance (Podsakoff et al., 2003). In addition, as a diagnostic step, we found that an unrotated factor solution of a Principle Factor Analysis did not result in one adequate single factor emerging from the analysis that accounts for the majority of the covariance between the variables. Factor loadings, uniqueness, and Kaiser–Meyer–Olkin measures are available from the first author upon request.

3.4 Statistical analysis

Since in our hypotheses time is a predictor variable, and we assume that individual and firm-level variables influence how turnover intent develops over time, growth modeling using random coefficient models is the most appropriate method for statistical analysis (Bliese & Ployhart, 2002; Schonfeld & Rindskopf, 2007). “Growth modeling involves looking at how individuals (or units, groups, organizations, etc.) change over time and whether there are differences in patterns of change” (Bliese & Ployhart, 2002, p. 363). In our study, we focus on how turnover intent of individuals in ventures develops over time, and whether individuals’ start-up work experience and venture growth moderate the change pattern. In addition to its analytical fit, growth modeling with a random coefficient model has the benefit that missing data do not compromise the analysis. In the case of our data collection, employees might have not participated in one round but resumed their participation in a later round. Also, some employees left and joined the ventures we examined over the course of the study. Therefore, our data do not represent a balanced panel of employees who participated in each round of data collection. However, it is “evident that one of the strengths of RCM [random coefficient models] is that the missing data pose no particular problems in terms of estimation. Basically, the parameter estimates are based on the available information” (Bliese & Ployhart, 2002, p. 365).

Finally, we note that our data set involves a nested structure based on levels of analysis and temporal aspects. We collected the data in four rounds (level 1) from employees (level 2) working in different ventures (level 3). We collected the criterion variable turnover intent in each round from the employees. Therefore, turnover intent is a longitudinal variable that varies over time. Our model assumes that employees’ start-up work experience will influence turnover intent over time. Start-up work experience reflects the number of years employees had worked in start-ups before joining the current venture. Therefore, this variable is constant for each employee over the observation period, and we modeled it as an individual-level constant variable. We modeled the individual-level control variables the same way. The model further assumes that venture growth over the observation period will impact the relationship between time and turnover intent. As this variable is the same for all employees of a venture, we modeled it as a firm-level constant variable. We modeled the firm-level control variables firm age, number of employees (t1), and environmental dynamism, the same way. Figure 1 displays the focal variables of this study and the level at which they reside. The time-varying variables represent the individual longitudinal level (level 1), the individual variables represent the individual constant level (level 2), and the firm variables represent the firm constant level (level 3). We used Stata version 16.1 for analyses.

Fig. 1
figure 1

Research model

Full size image

4 Results

As an initial step, we determined whether a growth model is the appropriate model for our data. We follow Bliese and Ployhart (2002) and compare various models of increasing complexity with respect to model fit. Model 1 includes the time variable as a predictor without adding any random components to the model; that is, neither a random intercept nor random slope is included. This model represents the baseline model. Model 2 is a random intercept model but does not include random slopes. Comparing the baseline model to model 2 (with random intercepts) reveals that model 2 fits the data significantly better than the baseline model (likelihood ratio = 437.91; p < 0.001). In model 3, a random slope for time is added to the random component of the model. A comparison of model 2 and model 3 reveals that model 3 fits the data significantly better (likelihood ratio = 43.95; p < 0.001). This supports the appropriateness of specifying a random intercept and random slope model for our data. Finally, we imposed restrictions on the residual errors to account for potential autocorrelation and heteroskedasticity. First, in model 4, we included a first-order autoregressive residual variance structure. When contrasting model 4 to model 3, results show that model 4 fits the data significantly better (likelihood ratio = 6.15; p < 0.05). Finally, we add restrictions to the residual errors to allow for heteroskedasticity with respect to the time variable. Model comparison results reveal that this model does not fit the data better (likelihood ratio = 2.00; p > 0.1). Although test results do not indicate issues due to heteroskedasticity, all models were estimated using robust standard errors.

4.1 Correlations and descriptive statistics

We display descriptive statistics and correlations between the focal study variables in Table 1. All VIFs were below 10 with the maximum value being 1.72, which is an acceptable threshold (Hair et al., 2013). This indicates that multicollinearity is unlikely to be a concern in our study.

Table 1 Means, standard deviations, and correlationsa
Full size table

4.2 Hierarchical linear modeling

Table 2 shows the results of the random coefficient growth modeling analyses. The first model is the null model, which we use to estimate the variance components. The intraclass correlation (ICC) denotes the proportion of total variance in the criterion variable that is related to each level of the model. Intraclass correlation calculations were based on Raudenbush and Bryk (2002). In this study, 40.1% of the variance in turnover intent resides at the longitudinal level (level 1), 46.8% within individuals (level 2), and 13.1% resides between firms (level 3). The theoretical model hypothesizes that variables at three levels—level 1 variables (time), level 2 variables (start-up work experience), and level 3 variables (venture growth)—conjointly impact employees’ turnover intent. Only a considerable amount of variance at each level justifies the use of a multilevel modeling approach (Aguinis et al., 2013). The results of the null model indicate that there is considerable variance at each level, which supports the use of a multilevel approach. Model 2 shows the results when we include only the control variables. Employee age, tenure, and employee education show a significant association with turnover intent (coefficient =  − 0.030, p < 0.05; coefficient = 0.157, p < 0.01; coefficient = 0.243, p < 0.05, respectively).

Table 2 Results of growth modeling using random coefficient model
Full size table

Model 3 displays the results for the predictor variable, time, which shows a significant positive association with turnover intent (coefficient = 0.263, p < 0.001). This result is consistent with Hypothesis 1; that is, in young ventures, employees’ turnover intent increases over time. Including time as a predictor variable explained 8% in turnover intent. The level 1 R2 calculation is based on Raudenbush and Bryk (2002). Also, as the research model assumes that the increase of turnover intent over time varies between individuals and firms, we estimate models with random slopes for the variable time. Overall, 78.3% of the variance in the relationship between time and turnover intent (i.e., the slope variance of time) is at the individual level (level 2) and 21.7% of the variance is at the firm level (level 3). Again, this calculation is based on Raudenbush and Bryk (2002).

In model 4, we add the main effect of start-up work experience, which does not show a significant direct relationship with employees’ turnover intent (coefficient =  − 0.060, p > 0.1). Model 5 includes the main effect of venture growth which does not reveal a significant relationship with turnover intent (coefficient = 0.010, p > 0.1). In model 6, we include all main effects of time (coefficient = 0.265, p < 0.001), start-up work experience (coefficient =  − 0.059, p > 0.1), and venture growth (coefficient = 0.010, p > 0.1).

Model 7 includes the interaction between time and start-up work experience, which is negatively and marginally significantly associated with employees’ turnover intent (coefficient =  − 0.022, p < 0.1). The interaction is depicted in Fig. 2. The y-axis represents turnover intent, the x-axis the independent variable time, and the graphs are relationships under the conditions of low or high start-up work experience (two standard deviations below and above the mean). Figure 2 shows that in young ventures, the increase of employees’ turnover intent over time is weaker for employees with longer work experience in start-ups prior to the current one, as predicted in Hypothesis 2. Moreover, simple slope analysis shows that there is a significant positive relationship between time and employees’ turnover intent when start-up work experience is low (simple slope = 0.327, p < 0.001) and when start-up work experience is high (simple slope = 0.203, p < 0.001).Footnote 1 The interaction between time and start-up work experience explains 6% of the slope variance at level 2 (explained slope variance was calculated based on Aguinis et al., 2013), corresponding to a weak interaction effect (e.g., Aguinis et al., 2005). In practical terms, the interaction plot reveals that over a period of 1 year, turnover intent for employees with lower start-up work experience increased by 34.71%. For employees with higher start-up work experience, turnover intent increased by 30.91%.

Fig. 2
figure 2

Interaction graph of time and start-up work experience

Full size image

In model 8, we include the interaction term between time and venture growth and find a negative and significant association (coefficient =  − 0.006, p < 0.01). Again, we depict the interaction graphs in Fig. 3, representing the relationships between time and turnover intent for ventures experiencing periods of low and high venture growth (two standard deviations below and above the mean). Consistent with Hypothesis 3, Fig. 3 shows that in young ventures, the increase of employees’ turnover intent over time is weaker for ventures experiencing higher growth. Moreover, simple slope analysis shows that there is a significant positive relationship between time and employees’ turnover intent when venture growth is low (simple slope = 0.322, p < 0.001) and when venture growth is high (simple slope = 0.234, p < 0.001). The inclusion of this interaction term explains 42.2% of the slope variance at level 3 (Aguinis et al., 2013), corresponding to a strong interaction effect. In practical terms, the interaction graph shows that over the period of 1 year, turnover intent of employees in ventures with lower growth increased by 42.79% compared to an increase of 29.16% in ventures with higher growth. Model 9 includes both interaction terms of start-up work experience and venture growth with time, respectively. Again, the results show a marginally significant and negative association between turnover intent and the interaction between time and start-up work experience (coefficient =  − 0.022, p < 0.1) and a negative and significant relationship between turnover intent and the interaction between time and venture growth (coefficient =  − 0.006, p < 0.01).

Fig. 3
figure 3

Interaction graph of time and venture growth

Full size image

Finally, model 10 includes the three-way interaction term between time, start-up work experience, and venture growth, and shows a significant positive relationship with turnover intent (coefficient = 0.003, p < 0.001). Figure 4 displays two sets of graphs for the condition of high and low start-up work experience. The graphs show that the increase of employees’ turnover intent over time is weaker in ventures experiencing relatively high growth, but this interaction effect is stronger for employees with less (shorter) pre-entry work experience in start-ups. Simple slope analysis shows that when pre-entry start-up work experience is low, there is a significant positive relationship between time and turnover intent both when venture growth is high (simple slope = 0.232, p < 0.001) and when venture growth is low (simple slope = 0.435, p < 0.001). When start-up work experience is high, there is a significant positive relationship between time and turnover intent, both when venture growth is high (simple slope = 0.237, p < 0.001) and when venture growth is low (simple slope = 0.209, p < 0.001). In addition, slope difference tests (Dawson & Richter, 2006) show that for low start-up work experience, the slopes for low and high venture growth differ significantly (t =  − 4.246, p < 0.001). However, for high start-up work experience, the slopes for low and high venture growth do not differ significantly (t = 1.578, p > 0.1). This provides support for Hypotheses 4. In practical terms, the interaction graphs reveal that for employees with lower start-up work experience, turnover intent increased by 58.18% in ventures with lower growth compared to an increase of 26.52% in ventures with higher growth rates over a 1-year time frame. For employees with higher start-up work experience, turnover intent increased by 27.57% in ventures with lower growth and by 32.32% in ventures with higher growth over 1 year.

Fig. 4
figure 4

Interaction graph of time, start-up work experience, and venture growth

Full size image

4.3 Robustness checks

We ran multiple robustness checks to corroborate our results. First, at the individual level, we also controlled for employee founding experience (measured as the number of ventures founded by the employee prior to joining the current venture), as individuals with founding experience may have more realistic expectations about start-up work environments. In addition, we controlled for ownership in the current venture (coded as 1 = ownership and 0 = no ownership) since ownership may decrease intent to leave the venture. As mentioned above, we asked respondents to report major events for the venture that had happened since the last survey round. Participants described 1,463 events in total. We coded the events into positive and negative events because the occurrence of such events might influence employee turnover intentions. Negative events included, for example, the loss of a major customer, the unsuccessful closing of a planned financing round, and failure of product development projects. Positive events included accomplishment the successful closing of a financing round, closing a substantial customer contract, and market launch of a new product. At the firm level, we additionally controlled for location (a categorical variable ranging from 1 to 3 reflecting the three major German metropolitan areas where data were collected) as well as the unemployment rate in the metropolitan area (indicating alternative job opportunities). We also used environmental hostility as an alternative control variable because in hostile environments, employees may find more job opportunities at competitors. Environmental hostility was measured using the 6-item scale by Green et al. (2008). Measurement items included “the failure rate of firms in my industry is high” or “competitive intensity is high in my industry.” Participants indicated their agreement with these statements on a 7-point Likert-type scale ranging from 1 = “strongly disagree” to 7 = “strongly agree.” Cronbach’s alpha was 0.69. In our survey, only founders assessed environmental hostility. When two or more founders participated in our study, we used the mean of their responses for the analysis. The variable was grand-mean centered. Moreover, as another alternative to capture firms’ environmental conditions, we included a variable reflecting the industry of the ventures with 1 referring to the venture being in the IT industry and 0 otherwise. Finally, we acknowledge the discussion about the use of control variables and that some authors advocate for the use of few or even no control variables (Carlson & Wu, 2012). Thus, we also estimated models without control variables. The results of our analyses remain stable and support our hypotheses across all alternative controls and model specifications.

To corroborate the results of the three-way interaction, we ran additional analyses. Specifically, we split the sample in two groups with one group consisting of employees with start-up work experience that was above our sample average and the other group consisting of employees with below sample average start-up work experience. For the subsample of above-average start-up work experience, the interaction between time and venture growth is marginally significant (coefficient =  − 0.004, p < 0.1). For the sample consisting of employees with below-average start-up work experience, the interaction between time and venture growth is significant (coefficient =  − 0.007, p < 0.001). In sum, this result corroborates our finding that for employees with above-average sample start-up work experience, turnover intent over time is not strongly influenced by venture growth. In contrast, for employees with below-average start-up work experience, the relationship between time and turnover intent is stronger in low growth ventures than in high growth ventures.

Finally, while we do not have information on who actually left the ventures examined in our sample during the observation period, we know who dropped out of the sample. Since all who left the venture also dropped out of the sample, we would expect that a positive correlation between turnover intent and dropping out of the sample indicates, at least to some extent, that high turnover intent also motivates employees in our sample to actually leave. To test this assumption, we created two groups of participants with one group including all participants who did not continue to respond to the survey in time 4 (potentially because they had left the firm) and the other group with all participants who continued to complete the surveys in time 4. We found that the group that continued to participate in time 4 had a significantly lower mean of turnover intent than the other group (t = 1.983, p < 0.05). Thus, our analysis suggests that higher turnover intent is related to participants’ dropping out from our sample, which provides at least some indication that higher turnover intent may also be related to participants actually leaving the venture.

5 Discussion

Almost two decades ago, in their seminal review of HR management research in small ventures, Cardon and Stevens (2004) conclude that employee retention is one of the areas that research has overlooked. To the best of our knowledge, this study is the first in the extant literature to explore the relationship between person and organization-level characteristics and employees’ turnover intent in young ventures over time. Specifically, we theorize and find that the increase in turnover intent over time is mitigated by employees’ prior start-up work experience in a young venture and their firms’ experience of relatively high growth. That is, based on a unique, hand-collected, and longitudinal data set including 1,151 questionnaire responses from 458 employees of 67 ventures over 1 year, we find that employees’ turnover intent increased with time and that this relationship is (conjointly) moderated by their prior start-up work experience and venture employee growth.

5.1 Theoretical implications

Met expectations theory (Porter & Steers, 1973) maintains that a means to reduce new employee turnover is to align their expectations with the reality of the actual job and job environment. Similarly, socialization models of turnover maintain that adjustment processes that clarify new hire expectations, aid learning about how to do the job and how to function in the social and cultural work environment, and that develop perceived organizational support and embeddedness, reduce turnover (Allen & Shanock, 2013; Bauer et al., 2007). Past research suggests that lowering unrealistic expectations of new employees by informing them about what to expect in terms of policies, practices, norms, and values contributed to job satisfaction and retention (Buckley et al., 2002; Taylor & Giannantonio, 1993). Delobbe et al. (2016) note the importance of communication between the employer and employee (i.e., leader-member exchange) to mitigate against the development of negative attitudes and turnover. Moreover, socialization tactics can help employees engage in proactive coping, where the employee identifies methods to prevent work stress (Ślebarska et al., 2019) and can dampen a “reality shock” and reduce the likelihood of newcomer attrition (Wang et al., 2017). However, in the young venture context, high levels of organizational change and a lack of HR practices make it less likely that unmet expectations are mitigated. Indeed, we demonstrate that in young ventures, employees’ turnover intent increases over time. Therefore, our study not only highlights the boundaries of applying current theory and empirical findings to employee staffing and management in the young venture context but also indicates opportunities for enhanced HR management approaches based on mitigating unmet expectations post-entry (e.g., onboarding processes that address how individuals can manage rapid change).

At the level of the individual employee, although only at marginal significance, we identify prior start-up work experience as a contingency helping to explain the differences between the young venture and established firm contexts with respect to employee turnover intentions. This finding suggests that prior start-up work experience provides employees with relevant knowledge about the typical venture work environment and an understanding of the potential frequency and magnitude of change that characterizes this environment. Such employees are likely to have more realistic expectations about what it means to work in a start-up and hence are less likely to form unmet expectations after organizational entry. We thus complement existing research by identifying who is most likely to develop turnover intent in a young venture. We also suggest that while many ventures may require overall high educational skills from their employees (Cardon, 2003) due to the multitude and complexity of tasks to be accomplished, what may be relevant when retaining these employees with the venture is their prior work experience in a start-up.

At the level of the organization, our study identifies venture growth as a contingency of the time-employee turnover relationship. Specifically, our theorizing suggests that the increase in employees’ turnover intent is stronger when the venture experiences a period of lower growth. Employees may perceive venture growth as a promise of future need fulfillment and opportunities to meet pre-entry expectations (e.g., professional development), which makes it more attractive for them to stay with the venture. However, relatively slow-growing ventures provide fewer need fulfillment opportunities and opportunities to meet pre-entry expectations (e.g., personal development opportunities and long-term employment perspectives for employees). Our study thus identifies the important role of a venture’s growth rate in retaining employees—even if it may not necessarily impact existing employees directly, growth appears to provide an important cue for employees. Under such conditions, employees seem more willing to accept the change that accompanies the maturation of young ventures. This is theoretically interesting because the venture growth literature has argued that employees impact growth (e.g., through their skills and knowledge, Rauch et al., 2005), but the reverse relationship of how growth impacts employees is rarely discussed. Together with prior work on employees influencing growth, our study suggests the interesting possibility of a spiraling relationship (e.g., Shepherd et al., 2010) such that employees with a lower turnover intent drive growth, and more growth further diminishes the negative effect of time on their turnover intent, and so on. Future research that uses longitudinal data capturing a longer time period than this study can test the presence of such turnover intent-venture growth spirals in young ventures.

We theorize that the effect of time is also contingent on the interaction of venture growth and employees’ start-up work experience, such that the interaction between time and venture growth is strongest for employees with little start-up work experience working in ventures experiencing periods of low growth. Hence, for theoretically understanding employees’ turnover intentions in a young venture, it is insufficient to only consider whether the changes the venture experiences over time are embedded in an overall positive development of the venture (high growth period) or in an overall negative development of the venture (low growth period). Rather, it appears that employees judge venture development differently based on the extent of their prior start-up experience such that those with longer experience are more likely to “ride out the storm” that growth dynamics pose to young ventures. That is, a theoretical understanding of employee turnover intent in young ventures does not only require that turnover intentions change over time, but also one that acknowledges that employees are heterogeneous in how they react to a venture’s growth.

Indeed, these findings shed further light on past research on venture growth and turnover. In particular, when restricting their analysis to a sample of start-up firms that survived 6 or more years and experienced growing staff complements, Gjerløv-Juel and Guenther (2019) report higher employment growth during the first year increased employee turnover in the long run. Our results and theorizing complement this study by suggesting that turnover results from an interaction between the passage of time, an individual’s pre-entry work experience, and firm growth. Interestingly, Gjerløv-Juel and Guenther’s (2019) findings regarding the relationship between growth and turnover did not hold for growth during the second and third year and this may be because in their first year of employment, employees gained start-up work experience (e.g., the employee realizes that it takes time to realize the potential benefits of early employee expansion) that mitigated the effects of their perceived unmet expectations.

5.2 Limitations

Since we used a repeated measure design, there might be order effects in our data. Therefore, other study designs (e.g., experimental designs) provide opportunities to further investigate the relationships in our model. However, given the time between measurements in our study (3 months), order effects are likely low. Furthermore, while our data show considerable variance over the observation period, the period of 1 year could still be considered relatively short. For example, employees with shorter pre-entry start-up work experience might recover from a reality shock when they find their initial expectations unmet; however, this process might take a longer period of time than captured in our study. Hence, future research might consider longer time frames. Moreover, the extent to which employees experience unmet expectations might depend not only on the length of their start-up work experience (as tested here) but also on the number of start-ups they have worked for and the industry of those start-ups. Future research can add to our understanding of the role of start-up work experience by providing additional nuanced insights on how pre-entry work experience can shape turnover intent. In addition, past research indicates that interpersonal relationships such as family bonds and friendship between founders and employees may have an important influence on turnover intentions (e.g., Francis & Sandberg, 2000; Mahto et al., 2020). While we cannot account for interpersonal relationships with our data set, future research could investigate their role in moderating the time and turnover intention relationship. Finally, although turnover intent is a strong indicator of employees’ attitudes toward their employing organization, it is not a measure of actual employee exit. For example, employees might develop turnover intent but stay until they have found an alternative job opportunity. It would be interesting to study whether such employees are more harmful for the ventures than those who actually left because uncommitted employees may consume the venture’s resources without providing performance benefits. Future studies, in particular those exploring a longer time frame, might also include turnover intent as a mediator rather than a dependent variable and investigate when it actually translates into exit from the venture.

5.3 Practical implications

Retaining their best talent is difficult for young ventures because their limited resources and smallness often make it impossible to pay the high salaries, to offer comprehensive benefits plans (e.g., with pension and disability programs) and sophisticated HRM practices, and to provide the established career paths offered by larger and more established organizations (e.g., Hornsby & Kuratko, 1990; Massey & Campbell, 2013; Williamson, 2000). Nonetheless, young venture managers should be aware that their employees’ turnover intent tends to increase over time, indicating the need to actively manage employees and communicate potential changes within their ventures openly as a means to maintain realistic expectations. Met expectations theory provides clear direction on how to retain employees. For instance, ventures can provide realistic pre-entry previews (i.e., set transparent expectations) to reduce initial expectations (Phillips, 1998; Premack & Wanous, 1985), provide post-hire realistic job previews to increase perceptions that the firm is honest (Earnest et al., 2011), or hire from within the organization when possible and hire using employee referrals because they are more likely to have received a realistic preview of the work environment (e.g., Schlachter & Pieper, 2019).

In addition, met expectations theory suggests that one way of reducing employee turnover intent is by ensuring the venture recruits the right employees in the first place. It appears that those applicants who have prior work experience in start-ups are those that will most likely “ride out potential storms” in a young venture. Managers are well advised to pay attention to prior start-up experience in applicants’ CVs; perhaps, in particular, if managers operate in high-risk firms entering volatile markets, where periods of low (negative) growth are likely to follow those of higher growth.

Rather than managing pre-entry expectations, another management approach may involve venture leaders influencing the recrafting of employee narratives as they experience organizational change. Shipp and Jansen (2011) state that individuals look to recraft their narratives about whether they fit with an organization during periods of change, such as the surprise when finding that the actual work environment differs from one’s expectations. In this process, a venture leader’s emphasis on a positive future work environment likely reduces turnover intent (e.g., by reducing the experience of cognitive dissonance between past expectations and current work experiences; Shipp & Jansen, 2011). Pairing a new hire with an experienced employee in a mentorship relationship may be one way of both imparting relevant young venture work experience to a new hire who is without such experience and also a redemption narrative—where the storyline begins poorly, but improves over time, e.g., negative experiences will be “worth it” when the venture succeeds. When ventures experience poor growth, recrafted narratives could focus on lessening the threat of unmet expectations (e.g., lessening the loss of resources) and ensuring that the aspect of personal identity that could be threatened (e.g., loss of autonomy) is nurtured in some other way. Such an approach to managing turnover stands in contrast to contemporary approaches directed at changing single or multiple organizational factors (e.g., incentive systems or job design) to reduce turnover intent.

5.4 Conclusion

Our study illustrates the uniqueness of the young venture context for employee turnover intent, and we are only now starting to understand the importance and implications of HR management practices. Our study also highlights that one liability of newness for a young venture is that they experience increasing turnover intent over time, and we argue that this may be because there is a greater likelihood of unmet expectations in young ventures due to rapid organizational changes. That is, employees who were initially attracted to a young venture become more likely to leave over time. We illustrate that venture growth and employees’ start-up work experience (contingent on the venture’s growth rate) moderate the relationship between time and turnover intent. Since employees are a key resource that is difficult to acquire and sustain for young ventures, we hope that our work inspires future studies on this important topic.