Path Dependence in New Ventures’ Capital Structures

Path Dependence Theory

The notion of organizational path dependence is frequently used in the field of management and organizational studies. It is founded on the work on technology trajectories identified by Arthur (1989) and David (1985). Generally speaking, path dependence is a dynamic theory that assumes that development trajectories are constrained by initial choices. While the path that will dominate an organization is not clear from the start, early decisions restrict what is achievable in the future, and thus, not all paths will be open to organizations. More formally, path dependence is defined as a process where an initial contingent event sets into motion patterns that reproduce more of the same decisions and actions, where after the range of choices successively are reduced until a potentially inefficient lock-in situation is reached (Sydow et al., 2009). Central to path dependence theory is that it clarifies the process of becoming persistent (Koch, 2011).

The path-dependent process can be divided into three stages, each governed by different regimes (Sydow et al., 2009). Phase I is built on contingent, i.e., nonpurposive and somewhat random, decisions or events, chosen from broad sets of possible opportunities. Having said that, “history matters” even for the initial decision or event. This means that the first choices have imprints from the past, where individuals make decisions based on their historic experiences, skills, and preferences (Karim & Mitchell, 2000; Sydow et al., 2009). While these early decisions have narrowing effects, considerable scope of alternative choices still remains in Phase I (Arthur, 1989; Sydow et al., 2009). Phase II begins with a critical juncture, meaning that a triggering decision or event sets off a self-reinforcing process that is hard to abandon. This process, in turn, consists of dynamic and repetitive self-reinforcing mechanisms that together increase the organization’s tendency to produce and reproduce more and more of the same activities or decisions. Sydow et al. (2009) highlight four central self-reinforcing mechanisms: coordination effects, complementarity effects, learning effects, and adaptive expectation effects. In addition, the formation of network ties is often appointed in the literature as another important self-reinforcing mechanism (Gulati, 1995; Hallen, 2008; Podolny, 1994). A self-reinforcing mechanism can directly support the selected path by providing short-term rewards (Vergne & Durand, 2011). It may also be negatively driven, discouraging the organization from entering less attractive paths. While alternative paths are still possible to pursue in Phase II, a dominant route has begun to emerge (Sydow et al., 2009). The transition to Phase III is characterized by further restrictions, leading to that the organization eventually is locked-in to a single state. That is, the dominant decision pattern has become fixed and deterministic, where actions to a large extent are bound to the path (Sydow et al., 2009). While lock-in does not automatically lead to failure and losses, the rigidity increases the likelihood of inefficiency (Sydow et al., 2009).

There are only a few studies, to our knowledge, that investigate how initial financing choices affect future capital structures in startup firms. Hirsch and Walz (2011) developed a theoretical model describing path dependence in early funding decisions made by startup firms. Vanacker et al. (2014), in a longitudinal case study, argued that early ties with VCs are path dependent. Three studies investigate book values of debt ratios from financial statements. Hanssens et al. (2016), by applying imprinting theory, show that early debt policies in startup firms determine future debt decisions. Hirsch and Walz (2019) observed mixed results when investigating the convergence of debt-equity structures across firms over time. Siqueira et al. (2018) combined imprinting and social entrepreneurship theories to explain the difference in leverage levels in for-profit social, versus commercial, enterprises. Like Vanacker et al. (2014) and Hirsch and Walz (2011), we consider path dependence theory to be particularly promising for explaining how early financing choices in new companies affect future capital structures.

Path Dependence in Startup Firms’ Funding Decisions

As discussed earlier, path dependence theory states that imprints from the past matters also for the initial decisions made in Phase I, where experience at the individual level is central for the formation of future paths (Karim & Mitchell, 2000; Sydow et al., 2009). This is considered to be particularly true in startup firms, where founder-CEOs play critical roles in setting the initial structure, strategy, and culture of their firms (Baron et al., 1999; Hanssens et al., 2016). This seems to also be the situation when it comes to funding choices. Traditional capital structure theories assume that managers have full access to, and are fully aware of, the entire funding universe when seeking financing, as funding is freely available to all firms and information about alternatives is broadly accessible in a perfect market (Eckhardt et al., 2006; Myers & Majluf, 1984). However, entrepreneurs typically lack detailed knowledge about available financing alternatives and are not in a position to find, evaluate, and negotiate with all possible finance providers (Seghers et al., 2012; Vanacker et al., 2014). Instead, few entrepreneurs seem to carry out an elaborate search for financiers at startup, but limit their search to a few potential capital providers that they know or have heard of before (Atherton, 2009; Cole, 2013; Vanacker et al., 2014). Hence, the founders’, especially the founder-CEO’s, prior experiences are likely to have a large impact on the initial choice of fund provider. We argue that this choice may be the triggering event in a path dependence process, representing a critical juncture that leads to a set of self-reinforcing mechanisms. Of the self-reinforcing mechanisms highlighted in the literature (Gulati, 1995; Hallen, 2008; Sydow et al., 2009), we identify three that seem to be particularly applicable when it comes to startup financing: (i) learning effects, (ii) network tie effects, and (iii) adaptive expectation effects.

Learning effects are central in the path dependence literature, referring to the rigid character of organizational competences and capabilities (Cohen, 2007; Miller, 1993). Learning effects imply that the more often a task is carried out, the more skillfully and efficiently it will be performed in the future (Sydow et al., 2009). While increased knowledge, as a result from learning, naturally has many positive effects, it may also hinder organizational development. This is because an organization that has developed successful strategies in the past tends to focus all learning on refining these skills rather than looking for alternative solutions (March, 2006). Supporting this, research on startups shows that entrepreneurial learning often demonstrates path-dependent features (Minniti & Bygrave, 2001), where resource-constrained entrepreneurs tend to revert to old strategies in change situations where new approaches instead are needed (Politis, 2005; Wright et al., 1998). That is, entrepreneurs primarily exploit opportunities related to information they already have, since exploring alternative paths require new knowledge that startups rarely have time to acquire (Choi & Shepherd, 2004; Shane, 2000). When seeking external funding, information and knowledge about the targeted type of financing are required, such as understanding who the potential financiers are, what interests them, how they work, and so on, (cf., Mason & Stark, 2004). Irrespective of being a provider of equity, a commercial bank, or a public authority offering subsidies, each group has its special requirements and expectations. Hence, we argue that learning about a specific type of financier will be important for successful fundraising. In line with this, previous research indicates that there is a learning effect from searching for financing (Vanacker & Manigart, 2010). However, due to funding sources’ complexity variations, learning effects are likely to differ. For example, seeking venture capital is considered more difficult than applying for bank debt (Baeyens & Manigart, 2006), likely due to that VCs put higher demand on business plans, engage in more extensive due diligence activities, and introduce more complex investor contracts (Kaplan & Strömberg, 2004; Sahlman, 1990). Following this, we propose that learning effects as a self-reinforcing mechanism leading to path dependence and potential lock-in situations are expected to be strong for firms utilizing external debt or subsidy funding, but even stronger for ventures capitalized by external equity.

Another self-reinforcing mechanism highlighted in the literature is the development of network ties, which we refer to as network tie effects. Organizations form ties with the same partners repeatedly and existing ties shape the formation of subsequent partnerships, leading to the evolution of networks becoming path-dependent (Gulati, 1995; Hallen, 2008; Hite & Hesterly, 2001; Podolny, 1994). An organization turns to existing partners not only to reduce search costs, but also for reducing the risk of opportunism (Williamson, 1985). This is particularly true in uncertain environments, which characterize the situation for most new ventures. Hence, startup firms are found to prefer relations with organizations where founders have either direct or indirect ties (Hallen, 2008; Manning & Sydow, 2011; Sullivan & Ford, 2014). This is also true for financial networks, albeit with some variance depending on the type of funder. Research concerning VCs and BAs shows that investment tie formation is path dependent (Hallen, 2008; Ibrahim, 2008; Vanacker et al., 2014). Not only does an entrepreneur prefer to approach a VC or a BA with whom she has prior relations (Ibrahim, 2008; Zhang et al., 2008), but these investors also tend to favor ventures founded by entrepreneurs they know (Hsu, 2007; Shane & Cable, 2002). Moreover, since VCs, and increasingly also BAs, typically syndicate their investments with other VCs or BAs, entrepreneurs get access to their investors’ networks (Brettel, 2003; Hallen, 2008; Mason et al., 2019; Meuleman et al., 2009). Taken together, entrepreneurs seeking equity funding typically get funded by investors with whom they have direct or indirect ties (Hsu, 2007; Shane & Stuart, 2002; Vanacker et al., 2014). Hence, we argue that self-reinforcing network tie effects to a very high extent contribute to the development of financing-related path dependence in startup firms funded by external equity.

It is likely that path dependence originating from network-related mechanisms also occurs when a startup is financed by external debt. Research shows that banking for decades has changed from a transaction- to a relationship-oriented business model (Berger & Dick, 2007; Dibb & Meadows, 2001). Following that, relationships with banks seem long-lasting since bank switching is rare (Durkin et al., 2013; Perry & Coetzer, 2009). Moreover, in contrast to large companies, small and young ventures are unlikely to deal with multiple banks (Howorth et al., 2003; Perry & Coetzer, 2009). Thereby, we argue that direct ties between startup firms and their external debt providers are strong and important, while indirect ties are not at play. Hence, we propose that the degree of path dependence stemming from network tie effects for startup firms utilizing external debt is high, albeit not as high as for firms funded by external equity.

When it comes to the provision of subsidies, nothing indicates that relations with authorities who provide public subsidies are particularly strong. Rather, the opposite seems to be the case, where employees at public authorities strive to hold an arm’s-length distance to the subsidy applicants (Söderblom et al., 2015). Further, there are no indications that public authorities seek to facilitate access to other subsidy, equity, or debt providers. Hence, we argue that the network-related path-dependence effect for subsidized firms is close to nonexistent.

A third important self-reinforcing mechanism highlighted in the path dependence literature is referred to as adaptive expectation effects (Schreyögg et al., 2011). These effects concern the interactive building of preferences among organizational actors, where preferences are not set from the start but from step by step in a self-reinforcing process (Sydow et al., 2009). The underlying thinking is that actors seek acceptance and social belonging, and therefore adopt practices that are not necessarily in the best interest for the organization, but normatively are the most accepted (Scott, 2008). This leads to the emergence of dominant practices and solutions that are believed to be supported by other members of the organization. One actor who has particularly large, both formal and symbolic, authority to dictate strategies and practices in an organization is the CEO (Bertrand & Schoar, 2003; Gupta, 1988). She has the power to promote certain organizational behaviors and to discourage others (Dutta et al., 2016). Related to this study, research shows that CEOs imprint the financial policies of their firms, regardless of being optimal or not (Bertrand & Schoar, 2003; Hanssens et al., 2016).

Some entrepreneurs have clear opinions about whether their companies should be funded by external equity (de Bettignies & Brander, 2007; Ueda, 2004). The ones who favor external equity appreciate the value-added services that VCs and BAs bring to table in the forms of business skills and relevant networks (Berger & Udell, 1998; Paul et al., 2007)—and not least the signaling value and personal prestige associated with being funded by such actors (Hsu, 2004; Pollock et al., 2010). Hence, a normative, but not always rational, preference for external equity arises among these entrepreneurs. Other entrepreneurs dislike equity investors and want to avoid VCs and BAs since they do not want intruding in the firm’s management or any dilution of ownership (Block et al., 2017; Hellmann & Puri, 2000). But while the choice to not seek equity funding may be a deliberate decision, the literature does not provide any evidence that entrepreneurs develop a similar normative preference for external debt or subsidy financing. Taken together, we argue that adaptive effects have some impact as a self-reinforcing mechanism leading to path dependence for equity-funded firms, but no effect for firms using external debt or subsidies.

Based on the theorizing above, we propose that startup firms are likely to choose the same type of funding as was chosen initially, whether the funding was dominated by subsidies, external debt, or external equity. That is to say, even when a new venture has developed to such an extent that other sources of funding should be available and potentially optimal for the company, we suggest that they tend to return to the same financier who has previously provided capital, or at least, continue to favor the same type of financing as used previously—and thereby that startup financing is path dependent. But, we also propose that there will be some variation in the strength of path dependence among funding types. We have presented three self-reinforcing mechanisms that seem to be particularly applicable when it comes to understand the possible path dependence patterns of startup financing: learning, network tie, and adaptive expectation effects, reinforcing each other in a spiraling character. We argue that learning effects are strong for firms utilizing subsidies or external debt and very strong for firms funded by external equity. We argue that network tie effects are very strong for firms funded by external equity investors, strong when funded by debt providers but nonexisting when it comes to subsidy funding. Finally, we argue that adaptive expectation effects impact firms funded by external equity but are not at play for firms using external debt or subsidies. Given the accumulated effects stemming from these three self-reinforcing mechanisms, we thereby suggest that path dependence effects are very strong in startup firms funded by external equity, strong when funded by external debt, and medium for firms funded by subsidies, as illustrated in Table 1.

OPEN IN VIEWER

Table 1

Self-Reinforcing Mechanisms Underpinning Path Dependence for Subsidies, External Debt, and External Equity.

	Subsidy	External Debt	External Equity
Self-reinforcing mechanisms
Learning effects	++	++	+++
Network tie effects		++	+++
Adaptive expectation effects			+
Path dependence	Medium	Strong	Very strong

This leads to our first set of hypotheses:

Change in Path Dependence in Startup Firms’ Funding Decisions

While path dependence has deterministic properties, it may at some point come to an end (Djelic & Quack, 2007; Sydow et al., 2009). This is particularly true for organizational path dependence, being more complex and ambiguous and thereby to some extent less restrictive than market or technology lock-ins (Sydow et al., 2009). Dissolution of paths can happen accidentally and unintentionally, for example if an exogenous shock unsettles the entire system (Tripsas & Gavetti, 2000). But breaking out of dependence can also be triggered by internal events—particularly in sensitive periods. Such mindful deviations from paths may be rare, but they do occur (Garud & Karøe, 2001). In most organizations path dependence processes commence in the first sensitive period of their life, such as at founding. However, organizations typically face several sensitive periods during their lifetime, when the chance of breaking an existing path is larger and a new path may be embarked (Simsek et al., 2015; Sydow et al., 2009). An example of such a sensitive period is when bringing in a new CEO.

Many organizations are characterized by a central authority and hierarchical control, where the top management, particularly the CEO, exerts great influence on planning procedures, formal rules, and incentive systems (Bertrand & Schoar, 2003; Gupta, 1988). Thus, by replacing key individuals with obsolete knowledge, networks, or values, new beliefs will be brought to the table that potentially could change the direction of the organization (Burgelman, 2002; Siggelkow, 2001). That is, a fresh view on existing behaviors and practices enables knowledgeable actors to reflect on current practices and potentially to undertake path-breaking activities (Sydow et al., 2009). Supporting this, empirical research shows that the appointment of a new CEO increases the likelihood of other more substantive changes within a firm (Beckman & Burton, 2008; Castaldi & Dosi, 2006; Decker & Mellewigt, 2012).

Earlier, we presented three self-reinforcing mechanisms that we considered to be particularly likely to explain the formation of path dependence in startup firms’ financing efforts. We also appointed one person as being central in all these processes: the founder-CEO. We proposed that she plays a salient role in the development of organizational learning, formation of network ties, and in building preferences, which may over time transform into persistent organizational routines and a potential path dependence. Hence, we argue, in line with the previous discussion, that changing CEO is likely to increase the chance of breaking a path—also when it comes to the choice of external funding.

This can be explained from two perspectives. First, a new CEO has experiences, networks, and skills that differ from the predecessor. A new CEO does not have to defend previous decisions and behaviors, but instead is expected to question existing practices and present new strategic directions (Beckman & Burton, 2008; Decker & Mellewigt, 2012). Consequently, a new CEO who recognizes weaknesses in a current regime and strategically acts to escape a lock-in situation is able to embark on a new path. Second, the departure of a CEO might be a result of a need for change identified by others, such as owners or the board of directors. Regardless of underlying reason, a CEO shift enables a new strategic perspective, whereby the commitment to a past strategic path can be overcome. In line with this, Hanssens et al. (2016) show that CEO turnover affects debt-equity ratios in startup firms. We expand their reasoning by suggesting that CEO turnover increases the likelihood of changing capital structures in new ventures for all forms of external funding, by hypothesizing:

Other sensitive periods emerge when organizations are forced to radically change strategic direction. That could be when the venture establishes a new core business, enters a new geographical market, abandons a business unit, or when facing a larger organizational restructuring—events that may trigger the striking of a new strategic path (Castaldi & Dosi, 2006; Decker & Mellewigt, 2012; Klepper & Simons, 2000). We argue that a radical change to organizational strategy enables embarking on new paths also for capital structures in new ventures. This leads to our final hypothesis:

Alternative “History Matter” Theories

It is important to emphasize that path dependence is not an umbrella concept for all theories that explain persisting practices or long-standing procedures in organizations (Schreyögg & Sydow, 2011; Vergne & Durand, 2011). To clarify the unique characteristics of path dependence theory, a brief presentation of overlaps and distinctions with three related theories—learning theory, transaction cost theory, and imprinting theory—is provided next.

Learning theory proposes that organizational capacities develop from repeated trials (Cohen, 2007). The more often an activity is undertaken, the faster and more efficient it will be performed in the future. The theory also suggests that learning may have disadvantages since firms tend to rely on historical knowledge instead of exploring new alternatives even when faced with environmental changes (Argote, 1999; March, 2006). Thus, learning is an important self-reinforcing mechanism (Sydow et al., 2009), as highlighted earlier. However, while learning effects occasionally can be the sole drivers of path dependence, they more often interplay with other self-reinforcing mechanisms (Schreyögg et al., 2011; Vergne & Durand, 2011). We acknowledge that in this paper.

Transaction cost theory concerns how a complex transaction between two or more exchange partners should be governed to be as efficient as possible (Williamson, 1975, 1985). The decision about how to manage a transaction depends on the assets’ characteristics. High asset specificity, i.e., assets that cannot easily be redeployed, tends to initiate a process of escalating commitment to a particular relationship, whereby a path is likely to develop that potentially leads to a lock-in (Williamson, 1975, 1985). Although transaction cost theory acknowledges that history matters, the core components of the process, i.e., the unfolding of self-reinforcing mechanisms and the emergence of lock-ins, are not conceptualized in this theory (Schreyögg & Sydow, 2011). In addition, only network tie effects are taken into account, while path dependence theory provides a richer description of several self-reinforcing mechanisms and their interactions.

Imprinting theory (Beckman & Burton, 2008; Stinchcombe, 1965) is sometimes equated with organizational path dependence. There are indeed clear similarities. The basic idea of imprinting is that founders incorporate historically important elements to their organizations at origin, which become routinized and influence the organization’s structure and behavior long after its foundation (Stinchcombe, 1965). Thereby, both theories highlight the important role of initial decisions and operational procedures as repositories of organizational history (Sydow et al., 2009). However, despite these similarities, the process of becoming path dependent is built on a different logic than imprinting. First, while the replicated pattern in imprinting theory is set up at the beginning of the process, an organizational path is not clear at the start but shaped by later processes. Second, while path dependence theory describes the process of path formation, imprinting assumes that it is only the initial conditions that explain subsequent behavior and hence ignores the later dynamics of the process (Schreyögg et al., 2011). Still, imprinting plays an important role in path dependence processes in explaining the restrictions, and the founders’ key roles, in the formation phase (Schreyögg et al., 2011).

Dependent Variables

The traditional method to investigate capital structures among startup firms is to analyze leverage ratios, measured as the value of debt over the value of shareholder equity taken from companies’ balance sheets (Hanssens et al., 2016; Hirsch & Walz, 2019; Siqueira et al., 2018). Such an approach makes it impossible to separate internal funding from financing arriving from external sources. Another issue with using book-value data from financial reports is that only debt and equity are studied, while neglecting an additional common financial source to startup firms—subsidies. Moreover, when using balance sheet data, transactions made within the same year cannot be traced. To overcome these issues, we introduced an extension of the ratio concept, measuring the value of new subsidies, external debt, or equity to the total capital obtained in a certain year. We gathered the information through our surveys, capturing new external funding in 2012/2013 and new external funding in 2016/2017. We explicitly asked for new funding, and thereby we were able to exclude follow-up investments, extensions of loans, and so on. This means that although a venture might have received finance from a person or an organization previously, the new funding that we capture is legally independent from any previous payments.

To assess to what extent a new venture’s capital structure changes or not across time, we proceeded as follows. We first captured the new external funding each firm received during 2012/2013 through our first survey, where we calculated the ratio of capital stemming from subsidies, external debt and external equity for the firm in question. For example, in case a firm received a total of 100 TSEK in new external funding in 2012/2013 whereof 30 TSEK arrived in the form of subsidies and 70 TSEK as loans, the subsidy ratio was calculated to 30% and the debt ratio to 70%. In our second survey, we followed the same principle to capture the ratio of external financing sources in 2016/2017. Thereby we arrived at our first three dependent variables, Subsidy ratio_16/17 , Debt ratio_16/17, and Equity ratio_16/17 , capturing the respective shares of each source in 2016/2017.

The final dependent variable, Funding change, measures change in financial structures. That is, it captures substantive change from one capital structure to another (1 = change, 0 = no change). For example, the variable was coded (0) for a firm that had a debt ratio of 70% and a subsidy ratio of 30% in 2012/2013 and the same structure in 2016/2017, but (1) if the firm changed to having a debt ratio of 30% and an equity ratio of 70% in 2016/2017. Firms that modified the ratios of subsidies, external debt, and external equity but where the same source continued to dominate were coded as unchanged.

Independent Variables

The three independent variables used for testing the first set of hypotheses capture the dominating funding source year 2012/2013. These are ultimately the same measures as our first three dependent variables described earlier but for 2012/2013, named Subsidy ratio_12/13 , Debt ratio_12/13, and Equity ratio_12/13 .

When testing the set of hypotheses regarding change in capital structure, we utilize two independent variables related to change in leadership and strategic direction. ³ Hypothesis 2 concerns whether a firm changed CEO between 2012/2013 and 2016/2017. The binary variable CEO change, captured from our second survey, was used for this hypothesis. Hypothesis 3 measures strategic change. We, therefore, include the independent variable Strategic change. This variable was collected from the second survey as well, from responses to the question: “Has the venture made any radical changes in the business during 2,012 to 2017,, for example, [..] organizational restructuring, changes in offering, development of new sales channels, etc.,?” (scale from 1 = not at all, to 6 = very large extent).

Control Variables

In order to examine the unique explanatory power of our path dependence hypothesis, we incorporated a series of control variables related to pecking order and tradeoff theories. The control variables are lagged 1 year in order to avoid problems of reverse causality. In line with the theories, available internal financing, debt capacity, financial distress, tax shields, and agency costs will determine companies’ use of external funding (Cole, 2013; Vanacker & Manigart, 2010).

While pecking order theory predicts that companies with more internal financing seek to avoid expensive external funding (Myers & Majluf, 1984), tradeoff theory suggests that firms with excessive internal funds are expected to rebalance their capital structures by adding more external debt (Miller, 1977). We capture internal financing through two profitability measures: (i) earnings before interest and taxes in thousands of SEK, EBIT _t−1 , and (ii) return on assets, calculated as net earnings on total assets, ROA _t−1 (Cole, 2013; Vanacker & Manigart, 2010). We also use a liquidity measure to capture internal financing, Cash & equivalents/total assets_t−1 (Vanacker & Manigart, 2010).

According to tradeoff theory, companies with debt capacity seek to attract more financial debt in order to optimize their capital structures (Miller, 1977), whereas pecking order theory suggests that firms with limited debt capacity are more likely to be funded by external equity than by external debt (Vanacker & Manigart, 2010). We capture debt capacity in two ways: (i) leverage, measured as the ratio between debt and total assets, Debt ratio_t−1 , where high levels indicate limited debt capacity, and (ii) internally generated cash flow, Operating CF/total assets_t−1 , reflecting higher debt capacity (Vanacker & Manigart, 2010).

Further, tradeoff theory suggests financial distress can offset the benefits of tax shields and thus cause higher levels of equity funding (Miller, 1977). Pecking order theory proposes that financial distress makes it disadvantageous or even impossible to raise external debt, forcing firms to turn to external equity (Myers & Majluf, 1984). We capture financial distress in three ways (Cole, 2013; Vanacker & Manigart, 2010): (i) firm size, as net revenues in thousands of SEK, Sales_t−1 ⁴ , (ii) bankruptcy risk, captured as cash and cash equivalents divided by current liabilities, Quick ratio_t−1 , and (iii) ratio of tangible assets, i.e., property, plant and equipment, of total assets, Tangible assets/total assets_t−1 .

Tax shields are central in tradeoff theory, stipulating that ventures with large tax shields require less debt since they have already relaxed their tax burden (DeAngelo & Masulis, 1980; Miller, 1977). Interest payments are debt tax shields that reduce taxable income. In addition, depreciation and other nondebt tax deductible items can serve as substitutes for tax shields. We proxy debt and nondebt tax shields through two variables: Interests/total assets_t−1 and Depreciations/total assets_t−1 (Vanacker & Manigart, 2010).

Agency costs are central in both traditional capital structure theories and considered particularly relevant for firms with growth opportunities. Since growth opportunities are difficult to value for loan providers, tradeoff theory predicts a negative relationship between growth opportunities and debt (Myers, 1977). In the same vein, pecking order theory suggests that firms with growth opportunities are more likely to be funded by external equity than by external debt due to information asymmetry (Myers & Majluf, 1984). As a proxy for growth options, we capture entrepreneurs’ expectations of employees in 5 years, Growth ambition, collected from survey responses.

We also included a set of general control variables. A contextual proxy is included based on the founding year of the firm, Firm founding year, measuring the venture’s year of incorporation, where negative coefficients are associated with younger firms. This variable also provides a control for macro-related factors since all ventures in a year-cohort are likely to be affected by the same contextual factors. We control for regional differences by capturing the largest regions with three dummy coded regions: Region Stockholm, Region Göteborg, and Region Skåne (with the rest as the hold-out group). In addition, we control for five types of industries: Retail, Health & education, IT & telecom, Corporate service,and Construction, as these are the largest industries in the Serrano database (with the rest as the hold-out group). ⁵

Table 2 summarizes the variables used in the study and Table 3 reports their correlations. Most independent and control variables exhibit small-to-moderate correlations (max <.521), suggesting that multicollinearity is not a major threat to our data (Tables 2 and 3) .

OPEN IN VIEWER

Table 2

Description of Dependent, Independent, and Control Variables.

Variables	Description	Mean	SD	Min	Max	N
Subsidy ratio16/17	Ratio of subsidies to total new external funding 2016/2017	.009	.088	0	1	1,756
Debt ratio16/17	Ratio of external debt to total new external funding 2016/2017	.113	.310	0	1	1,756
Equity ratio16/17	Ratio of external equity to total new external funding 2016/2017	.019	.128	0	1	1,756
Funding change	Binary, “1” for firms that changed capital structure from 2012/2013 and 2016/2017, “0” otherwise	.022	.148	0	1	1,756
Subsidy ratio12/13	Ratio of subsidy to total new external funding 2012/2013	.025	.145	0	1	1,756
Debt ratio12/13	Ratio of external debt to total new external funding 2012/2013	.235	.419	0	1	1,756
Equity ratio12/13	Ratio of external equity to total new external funding 2012/2013	.029	.159	0	1	1,756
CEO change	Binary, “1” change of CEO between 2012/2013 and 2016/2017, “0” otherwise	.159	.366	0	1	454
Strategic change	Scale 1–6, strategic change based on respondents’ answers	3.035	1.754	1	6	452
EBIT t−1	Earnings before interest and taxEBIT, tSEK	740.759	4591.741	−52,006	77,291	1,756
ROA t−1	Return on assets, i.e., net earnings on total assets	.105	.423	−4	8	1,756
Cash and equivalent/total assets t−1	Cash and cash equivalents on total assets	.236	.226	0	1	1,756
Debt ratio t−1	Total debt on total assets	3.586	6.977	0	110	1,756
Operating CF/total assets t−1	Cash flow from operating activities, i.e., excluding investing and financing cashflows, on total assets	.089	.897	-5	13	1,756
Sales t−1	Net revenues, tSEK	16025.153	4385.141	0	995,254	1,756
Quick ratio t−1	Cash and cash equivalents on current liabilities	1.948	3.245	0	56	1,756
Tangible assets/total assets t−1	Tangible assets, i.e., property, plant, and equipment, on total assets	.272	.275	0	1	1,756
Interest/total assets t−1	Interests on total assets	.020	.057	0	1	1,756
Depreciations/total assets t−1	Depreciations on total assets	.052	.081	0	1	1,756
Growth ambition	Expected number of full-time employees in 5 years	22.004	43.222	0	1,100	1,756
Firm founding year	Year of founding	2,007.805	2.510	2,004	2,012	1,756
Region Stockholm	Binary, “1” Stockholm, “0” otherwise	.271	.445	0	1	1,756
Region Göteborg	Binary, “1” Göteborg, “0” otherwise	.149	.356	0	1	1,756
Region Skåne	Binary, “1” Skåne, “0” otherwise	.115	.320	0	1	1,756
Retail	Binary, “1” for retail industry, “0” otherwise	.198	.398	0	1	1,756
Health and Education	Binary, “1” for health and education industry, “0” otherwise	.096	.295	0	1	1,756
IT and telecom	Binary, “1” for IT, telecom, or software industry, “0” otherwise	.074	.261	0	1	1,756
Corporate service	Binary, “1” for corporate services industry, “0” otherwise	.308	.462	0	1	1,756
Construction	Binary, “1” for construction industry, “0” otherwise	.099	.299	0	1	1,756

OPEN IN VIEWER

Table 3

Correlations Among Variables in Study.

Variables		1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28
1	Subsidy ratio16/17
2	Debt ratio16/17	−.003
3	Equity ratio16/17	−.002	.000
4	Funding change	.033	.186**	.353**
5	Subsidy ratio12/13	.187**	.005	.025	.162**
6	Debt ratio12/13	.034	.188**	−.020	.047	−.055*
7	Equity ratio12/13	−.013	−.015	.279**	.415**	.011	−.072**
8	CEO change	.029	.013	.246**	.175**	.015	−.014	.194**
9	Strategic change	.069	.171**	.134**	.104*	−.032	.095*	.141**	.115*
10	EBIT t−1	.003	−.009	−.150**	−.101**	−.025	.028	−.131**	.064	−.018
11	ROA t−1	−.004	−.051	−.124**	−.167**	−.062*	−.012	−.136**	−.097*	−.085	.247**
12	Cash and equivalent/total assets t−1	.016	−.168**	−.011	−.023	.052	−.154**	−.014	−.002	−.201**	.049	.150**
13	Debt ratiot−1	−.009	.025	.013	.003	−.021	.044	−.021	.011	.022	−.057*	−.166**	−.221**
14	Operating CF/total assets t−1	−.020	−.054*	−.048	−.069*	−.042	.023	−.053*	.000	.033	.003	.181**	.007	−.078**
15	Sales t−1	−.014	.011	−.001	.052	−.010	.043	.098**	.132**	.041	.521**	.044	−.061*	.013	−.057*
16	Quick ratio t−1	−.005	−.060*	−.003	.007	.036	−.071**	−.004	−.016	−.133**	−.013	.066*	.270**	−.151**	.010	−.064*
17	Tangible assets/total assets t−1	.004	.115**	.077**	.064*	−.018	.111**	.064*	.003	.018	−.133**	−.139**	−.374**	.097**	−.030	−.050	−.097**
18	Interest/total assets t−1	−.013	−.006	.017	.022	−.031	.023	−.018	−.016	.118*	−.068*	.085**	−.164**	.089**	.264**	−.054*	−.087**	.085**
19	Depreciations/total assets t−1	.012	.042	.037	.018	−.016	.071**	.063*	.037	.072	−.154**	−.275**	−.125**	.056*	.053*	−.059*	−.114**	.366**	.171**
20	Growth ambition	−.016	−.011	.017	.044	−.026	.027	.149**	.121*	.083	.067*	−.009	−.022	.027	−.063*	.452**	−.042	.017	−.022	.000
21	Firm founding year	−.036	−.032	−.007	−.016	−.048	.047	.011	−.010	−.052	−.049	−.043	.067*	.091**	.028	−.075**	−.050	.066*	−.014	.190**	−.045
22	Region Stockholm	−.017	−.067*	.022	.058	.001	−.041	.087**	.020	−.015	.008	.006	.020	−.013	.033	.083**	.041	−.081**	−.036	−.021	.093**	−.057*
23	Region Göteborg	.010	.043	.029	−.008	.004	.036	.016	−.003	−.023	−.007	.024	.017	−.006	−.012	−.013	.042	.014	.046	.015	−.015	−.035	−.255**
24	Region Skåne	−.012	−.002	.016	−.014	−.030	.007	−.011	−.071	.091	−.023	−.028	−.015	.006	−.020	−.003	.011	.018	−.002	.007	−.004	.064*	−.220**	−.151**
25	Retail	−.029	−.039	−.053*	−.025	−.036	−.019	−.047	−.049	.079	−.022	.013	−.022	.038	.020	−.039	−.053	.005	.026	.056*	−.027	.034	−.050	.011	.014
26	Health and Education	−.028	−.038	−.030	.020	.002	−.003	.015	−.022	.021	.036	.029	.201**	−.054	−.047	.044	.056*	.001	−.039	.009	.162**	.069**	.033	.001	−.003	−.162**
27	IT and telecom	.002	−.047	.146**	.081**	−.013	−.065*	.137**	.182**	.077	−.059*	−.072**	−.007	−.053	−.010	−.031	.059*	−.023	−.016	.025	.037	−.077**	.077**	−.033	.019	−.140**	−.092**
28	Corporate service	.085**	−.046	−.021	−.026	.062*	−.021	−.009	.017	−.052	.037	.028	.045	−.059*	.076**	−.017	.040	−.050	.029	−.049	−.036	−.065*	.103**	.014	−.035	−.331**	−.217**	−.188**
29	Construction	−.029	.092**	.007	−.010	−.036	.042	−.032	−.098*	.007	−.001	.013	−.060*	.047	−.010	.014	−.017	−.029	−.022	−.055*	−.033	.081**	−.056*	.003	−.030	−.165**	−.108**	−.094**	−.222**

Notes. Significance levels: ^*** p < .001; ^** p < .01; ^* p < .05 (two-tailed). EBI = earnings before interest and tax; ROA = return on assets.