Abstract
Using country-sector data for 24 European countries from 2008 to 2019, we examine how economic complexity shapes firm dynamics. We study whether more complex productive structures foster entrepreneurship or instead act as a barrier to market entry, focusing on firm entry, exit, churn, and survival across size classes. To address endogeneity, we implement an instrumental variables strategy based on regional leave-one-out averages of the Economic Complexity Index (ECI). Our results suggest that higher economic complexity significantly reduces firm entry and business churn, and lowers medium- and long-term survival, particularly for larger firms. For microenterprises, the negative effects on longer-term survival are weaker, despite a stronger negative effect at entry. These findings suggest that economic complexity operates less as a broadly enabling entrepreneurial environment and more as a selective filter, with heterogeneous effects across firm sizes.
Similar content being viewed by others
1 Introduction
Firm entry, exit, and survival are fundamental to resource reallocation, productivity growth, and structural transformation (Bartelsman et al. 2013; Hsieh and Klenow 2014). Yet these processes do not occur in a homogeneous environment. Firms operate within productive structures that differ markedly in their technological and organizational sophistication, the density of complementary capabilities, and the extent to which production requires coordination across specialized activities. The Economic Complexity Index (ECI) summarizes these features by capturing the capability intensity embedded in an economy’s productive structure (Balland et al. 2022; Hidalgo and Hausmann 2009). While a large literature links economic complexity to aggregate growth, diversification, and upgrading, much less is known about how complexity shapes firm demographic outcomes.
This paper asks whether economic complexity acts primarily as a fertile environment or as a selective filter for firm dynamism. On the one hand, more complex economies may support entrepreneurship by fostering knowledge spillovers, specialization, and complementarities across firms and sectors (Audretsch and Feldman 1996; Balland et al. 2020; Boschma 2005). On the other hand, complexity may raise the capability threshold required for entry and survival. The literature on national and sectoral systems of innovation has long emphasized that firm performance depends on the density and quality of interactions among firms, universities, public agencies, and financial institutions, as well as on sector-specific knowledge bases and technological regimes (Lundvall 2010; Malerba 2004). Producing in capability-dense environments often requires tacit knowledge, non-trivial coordination, specialized routines, and sunk investments that not all firms can meet (Hidalgo 2015; Nelson and Winter 1982; Sutton 1991). Whether complexity ultimately stimulates dynamism or filters participation is therefore an empirical question.
We argue that, in advanced European economies, economic complexity operates primarily through a selective-filter mechanism. The idea is that high-complexity environments raise barriers to entry and reduce turnover by making participation more capability-intensive. At the same time, these effects need not be uniform across firms. Larger firms may be more exposed to adaptation pressures and frontier competition, whereas microenterprises may survive by concentrating in localized or less capability-intensive niches. This perspective implies that complexity can simultaneously discourage entry and generate heterogeneous survival patterns across firm size classes.
To examine this hypothesis, we assemble a panel of 24 European economies from 2008 to 2019 and estimate the relationship between local economic complexity and firm dynamics. Our outcomes include birth rates, death rates, business churn, and survival rates at different horizons. Because firm dynamics and complexity may be jointly determined, we complement fixed-effects estimates with a two-stage least-squares strategy that instruments domestic ECI using the leave-one-out average complexity of regional peers. This approach exploits regional technological co-movement while reducing the scope for reverse causality from domestic firm dynamics to measured complexity.
The results point to a clear pattern. Higher economic complexity is associated with lower firm entry and lower business churn. These findings are consistent with the view that complexity raises participation thresholds and dampens broad-based market turnover. The heterogeneity results further show that the negative entry effect is stronger for small firms. At the same time, the adverse association between complexity and survival is weaker for small firms than for larger firms, especially at longer horizons. Taken together, the evidence suggests that complexity does not operate as a uniformly entrepreneurial environment. Instead, it appears to restrict entry while producing uneven survival effects across firm types.
This paper makes two main contributions. First, we contribute to the literature on structural productivity and micro-mechanisms of selection (Haltiwanger et al. 2004; Restuccia and Rogerson 2017) by arguing that high-ECI environments may induce a form of lifecycle misallocation, in which firm survival becomes partially decoupled from growth potential. In doing so, it shows that the consequences of complexity depend not only on how much productive knowledge an economy embodies, but also on how that knowledge shapes selection into and within markets. Second, it contributes to the literature on firm dynamics and resource allocation by showing that structurally advanced environments may combine lower entry with heterogeneous post-entry persistence, particularly across firm size classes. These patterns are relevant for understanding how more complex productive structures interact with entrepreneurial renewal.
The findings also carry broader policy implications. If complexity raises capability barriers, policies that promote entry alone may be insufficient to generate inclusive business dynamism. In high-complexity environments, industrial policy may also need to expand access to complementary capabilities, specialized knowledge, and organizational support, particularly for smaller entrants facing higher barriers to entry.
The remainder of the paper proceeds as follows. Section 2 presents the theoretical framework. Sections 3.1 and 3.2 describe the data and empirical strategy. Section 4 reports the main results and robustness analyses. Section 5 discusses the interpretation of the findings. Section 6 concludes.
2 Theoretical framework
Firm entry, exit, and survival are components of economic development because they determine how productive resources are allocated across firms and sectors (Bartelsman et al. 2013). Persistent differences in aggregate productivity and income across countries have been linked to distortions in this reallocation process (Hsieh and Klenow 2009, 2014). These dynamics are driven by creative destruction (Schumpeter 1934, 1942), whereby innovation generates the turnover of firms and realigns economic activities toward more productive configurations (Aghion et al. 2014; Caballero 2010). The efficiency of this process depends on the institutional and organizational environment in which firms operate (North 1990; Restuccia and Rogerson 2017) and on the inherent distribution of firms’ capabilities (Geroski 1995; Nelson and Winter 1982).
If creative destruction is the mechanism that reallocates productive resources, then the complexity of the productive structure defines the environment in which selection and turnover occur. Complex economies are characterized by dense networks of specialized capabilities, interdependent knowledge flows, and high requirements for technological and organizational sophistication (Balland et al. 2022; Hidalgo and Hausmann 2009). The Economic Complexity Index (ECI) has been proposed as a summary measure of this embedded knowledge structure and capability distribution.
Capabilities are not directly observable, and the ECI should therefore be interpreted as an empirical proxy constructed from observed export patterns rather than as a direct measurement of an economy’s underlying knowledge stock. For example, recent work has shown that the index can also be understood as a low-dimensional summary of countries’ specialization patterns and similarities in export space, rather than as a literal readout of latent capabilities (Mealy et al. 2019). Moreover, because ECI is typically computed from gross export data, it may partly reflect participation in global value chains rather than only domestic productive knowledge (Koch 2021). At the same time, recent theoretical work provides a formal rationale for interpreting ECI as informative about productive capabilities, showing that, in capability-based production models, ECI can recover information about economies’ average capability endowments from specialization matrices (Hidalgo and Stojkoski 2025).
Mathematically, ECI is an eigenvector-based measure derived from the country-product specialization network. It is dimensionless, and its sign is conventional, so its interpretation does not come from an external cardinal scale of sophistication. Rather, within the economic complexity algorithm, products receive higher complexity scores when relatively few countries export them and, crucially, by countries that are themselves diversified in their export baskets. In this recursive sense, product complexity is defined endogenously by the structure of the country-product network: complex products are those that are non-ubiquitous but associated with diversified economies. A higher ECI, therefore, indicates that a country is positioned in the part of the export-specialization network associated with less ubiquitous products, interpreted in this literature as requiring broader sets of capabilities (Balland et al. 2022; Hausmann and Hidalgo 2011; Hidalgo and Hausmann 2009).
Within such environments, the dynamics of firm entry and survival may operate under two theoretically distinct regimes. On the one hand, complexity may serve as a fertile ground, enhancing spillovers, input sharing, and opportunities for specialization (Audretsch and Feldman 1996; Balland et al. 2020; Boschma 2005; Marshall 1890). On the other, it may act as a selective filter by increasing capability thresholds for participation (Cohen and Levinthal 1990; Nelson and Winter 1982; Sutton 1991), sustaining firms only if they can continuously adapt to evolving technological and competitive demands (Aghion et al. 2014; Dosi 2023; Hidalgo 2015).
The extent to which complexity fosters entry or imposes capability constraints is not uniform across firms. Differences in size, scope, and strategic positioning condition how firms experience these pressures (Acs and Audretsch 1990; Beck and Demirguc-Kunt 2006). Microenterprises may avoid direct exposure to frontier competition by occupying niche segments of localized demand (Duranton and Puga 2004; Moretti 2004), whereas larger firms, more tightly integrated in capability-dense production networks, may face higher adaptation burdens (Akcigit and Ates 2023). These asymmetries introduce the possibility of “entrepreneurial conservatism”, where smaller firms survive by deliberately avoiding growth or technological engagement (Coad 2009; Nightingale and Coad 2014), and “lifecycle misallocation”, where survival outcomes are driven not necessarily by productivity potential, but by capability alignment relative to environmental complexity (Hsieh and Klenow 2014; Restuccia and Rogerson 2017).
The following subsections articulate these mechanisms more precisely. Section 2.1 discusses complexity as a fertile ground that enables spillovers and innovation complementarities. Section 2.2 formalizes the view of complexity as a selective filter imposing capability thresholds on firms. Section 2.3 examines how these effects vary across firm size classes and capability profiles.
2.1 Economic complexity as a ‘fertile ground’
High economic complexity can function as a “fertile ground” by providing a dense ecosystem of knowledge, capabilities, and market opportunities, echoing Marshall’s insights on industrial districts (Marshall 1890). In contemporary economic complexity theory, ECI reflects the embedded collective know-how of an economy and its capacity to transform information into productive output (Balland et al. 2022; Hausmann and Hidalgo 2011; Hidalgo 2015). Within an effective Holistic Innovation System (Du and O’Connor 2021), higher complexity is expected to translate into greater innovation potential and a broader opportunity set for firms.
This fertile environment operates through three interrelated channels. First, dense networks of firms, institutions, and individuals facilitate knowledge spillovers and recombination of specialized capabilities and “personbytes” (Audretsch and Feldman 1996; Boschma 2005; Hidalgo 2015). Firms with sufficient absorptive capacity (Cohen and Levinthal 1990) can leverage these spillovers to access complementary knowledge, consistent with evolutionary learning frameworks (Dosi 2023; Nelson and Winter 1982). However, SMEs’ ability to benefit from such spillovers depends on cognitive proximity and network embeddedness (Boschma 2005).
Second, complex economies provide deeper markets for specialized inputs, skilled labor, and advanced business services (Balland et al. 2020). While this expands access to complementary assets, SMEs may still face constraints related to scale, bargaining power, and search friction in mobilizing those resources effectively. For example, consistent with the idea of a deep division of knowledge, large incumbents are more likely to internalize or routinely coordinate advanced legal, engineering, or R&D services. In contrast, micro and small firms may lack the organizational capacity required to orchestrate such fragmented inputs, even when they are locally abundant. As a result, dense capability markets do not necessarily translate into symmetric access, reinforcing size-dependent frictions within highly complex economies (Balland et al. 2020).
Third, the inherent diversity of high-ECI economies, captured by the structure of the product space (Hidalgo et al. 2007; Hidalgo and Hausmann 2009), generates multiple adjacent-possibility pathways for related diversification (Frenken et al. 2007; Neffke et al. 2011). These niches create strategic room for specialized entrants and for firms pursuing “entrepreneurial conservatism” by occupying defensible sub-segments rather than engaging directly with core complexity-intensive activities (Delgado et al. 2010). However, these adjacent possibilities are not uniformly accessible. In highly complex environments, diversification opportunities are increasingly conditioned on firms’ existing capabilities, scale, and organizational coordination, such that economic complexity functions as a selective filter that amplifies survival differences across firm types.
2.2 Economic complexity as a ‘selective filter’
While high-ECI economies are characterized by a rich set of diversification opportunities and dense capability markets, these features do not translate into an even playing field for all firms. Instead, increasing economic complexity systematically raises the minimum organizational, cognitive, and financial requirements needed to access, combine, and sustain productive activities. Paradoxically, the same complexity that generates opportunities imposes structural barriers that function as a selective filter for market participation. This mechanism can be conceptualized through the lens of endogenous sunk costs as in Sutton (1991). Competing in high-ECI environments, often associated with sophisticated sectors such as pharmaceuticals, advanced manufacturing, or ICT (Balland et al. 2022; Hausmann 2014), requires substantial upfront commitments in specialized human capital, R& D capabilities, and access to advanced supply chains. Beyond financial constraints, these commitments reflect a high threshold of absorptive capacity necessary to decode tacit knowledge embedded in dense capability networks (Cohen and Levinthal 1990; Hidalgo 2015).
This creates a rigorous selection mechanism at entry. The cumulative and capability-dependent nature of knowledge implies that entrants face a steep learning curve to establish operational credibility and legitimacy within the industry (Geroski 1995). Firms lacking either initial capability alignment or the ability to accumulate capabilities rapidly are filtered out through productivity-based selection, consistent with canonical industry dynamics models (Hopenhayn 1992; Jovanovic 1982).
Importantly, this filter is dynamic. High-ECI environments are characterized by rapid technological cycles and continuous processes of creative destruction (Caballero 2010; Schumpeter 1942) that shift the competitive frontier and redefine capability requirements for survival. For incumbents operating near this frontier, survival relies on permanent, resource-intensive adaptation to evolving technologies and market pressures (Aghion et al. 2014). Therefore, complexity imposes a dual pressure structure: a barrier to entry for new firms with limited resources and a continuous maintenance cost for incumbents subject to competitive churning and disruptive innovation (Ericson and Pakes 1995; Klepper 1996).
This adaptation burden is particularly relevant for firms operating close to the technological frontier. Innovation near the frontier is characterized by substantial uncertainty and by the need to recombine specialized knowledge across organizational boundaries. Recent discussions of biotechnology and artificial intelligence show that, in uncertain technological environments, the division of knowledge may precede the division of labor: firms first need to access, organize, and recombine specialized knowledge before defining the tasks and organizational arrangements through which innovation is carried out (Borsato and Llerena 2025, 2026). This mechanism is especially relevant for larger and more established firms. Although they possess greater resources, their established routines and internal specialization patterns may constrain search and adaptation when technological trajectories shift, making complexity a source of organizational inertia rather than only a source of opportunity.
In this sense, complexity does not merely sort firms at the moment of entry but continually reshapes the selection function as productivity thresholds evolve with technological trajectories. Firms remain viable only if they maintain capability coherence with the advancing frontier.
2.3 Heterogeneity across firms
The interaction between firms and high-ECI environments is marked by pronounced heterogeneity, as both the structural barriers of the selective filter and the opportunities of the fertile ground are experienced differently by firms depending on their size, capabilities, and sectoral embedding. The selective filter is more stringent for smaller firms. The intrinsic disadvantages associated with the “liability of smallness” (Cefis et al. 2022) are intensified by the burden of regulatory compliance, which often imposes disproportionate relative costs on SMEs (OECD 2021). Limited access to finance and constrained ability to invest in R&D or capability-building (Bui et al. 2021; Wang 2016) render the high fixed costs of participation (Sutton 1991) more prohibitive for young and small firms. Moreover, capability gaps in attracting specialized human capital, developing absorptive capacity (Cohen and Levinthal 1990), and integrating into collaborative networks (Boschma 2005; Freitas et al. 2024) restrict the ability of SMEs to navigate dense knowledge environments. This aligns with the empirical observation that higher ECI disproportionately reduces birth rates among microenterprises.
Paradoxically, those microenterprises that do enter tend to exhibit notable initial resilience. This suggests that only the most capable or strategically positioned ventures successfully pass the entry filter. Once established, such firms may exploit fertile-ground dynamics suited to their scale, operating within specialized market interstices (Coad 2009) or benefit from localized, relational knowledge spillovers (Audretsch and Feldman 1996). This implies underlying heterogeneity within the SME segment, with opportunity-driven and capability-rich firms more likely to survive early selection pressures (Storey 1994).
This initial resilience may extend to long-term persistence. Surviving SMEs often adopt an “entrepreneurial conservatism,” prioritizing stability and incremental adjustments over aggressive innovation or rapid expansion (Coad 2009). By operating in defensible niches, they insulate themselves from the escalating innovation requirements and competitive turbulence typical of firms positioned near the complexity frontier. Larger firms, by contrast, may be less adaptable and more constrained by bureaucratic inertia and rigid routines that slow capability adjustment (Coad 2009; Dranove et al. 2016). In such cases, a highly complex environment may prove more challenging for large incumbents than for strategically conservative small firms.
The fertile-ground aspects of high ECI, such as knowledge spillovers, talent mobility, and specialized input markets, may particularly benefit knowledge-intensive business services and technology-based SMEs (Du and O’Connor 2021). In contrast, SMEs operating in local, non-tradable services may primarily experience complexity through higher cost pressures, with their main benefit coming from increased aggregate demand rather than direct participation in advanced capability networks.
This multifaceted heterogeneity raises broader questions about lifecycle misallocation. If firms best adapted to persist in high-ECI contexts are not those with the highest potential for productivity gains, the Schumpeterian process of selection and reallocation (Hsieh and Klenow 2009; Restuccia and Rogerson 2017) may be muted or distorted. Consequently, the firms that remain may be stable but not growth-driving, potentially constraining aggregate dynamism in complex economies. The differential capacity of firms to absorb shocks, leverage spillovers, and adapt capabilities ultimately shapes the diverse trajectories of entry, survival, and growth within complex productive structures.
3 Materials and methods
This section outlines the data sources, variable construction, and empirical strategy employed to analyze how economic complexity influences firm dynamics. We begin by describing the data sources and defining the key variables used in our analysis. Following this, we outline the empirical strategy, including the econometric models specified and the estimation techniques utilized, with particular attention to addressing potential endogeneity concerns.
3.1 Data sources and definitions of variables
We used data on firm dynamics from 24Footnote 1 European countries between 2008 and 2019, with each industry aggregated at the NACE Rev.2Footnote 2 level, provided by Eurostat (OECD and Statistical Office of the European Communities 2008). Our dependent variables are the birth rate (%), death rate (%), business churn (%), survival rate 1 (%), survival rate 3 (%), and survival rate 5 (%). The birth rate and death rate are calculated as the rates of enterprise births and deaths in period t relative to the stock of active enterprises in t-1. The business churn rate is the sum of the birth and death rates for period t. Survival rates are calculated following the standard definition: the number of enterprises newly born in year xx-n that survived to year xx, expressed as a percentage of all enterprises born in year xx-n. In our analysis, we utilize 1-, 3-, and 5-year survival rates. Economic complexity was obtained at the aggregate country level, in its version calculated by HS6 export data by the Observatory of Economic Complexity (Stojkoski et al. 2023).
As shown in Table 7 (see appendix A), we employed a series of control variables commonly found in the firm dynamics literature, including total tax and contribution rate (Djankov et al. 2002, 2010; Sutton 1991; Tomasi et al. 2023), starting a business minimum capital (Ciccone and Papaioannou 2007; Djankov et al. 2002; Tomasi et al. 2023), human capital (Audretsch 2007; Chowdhury et al. 2019), domestic credit to the private sector (Rajan and Zingales 1998; Tomasi et al. 2023; Wang 2016), Economic Freedom of the World (Bjørnskov and Foss 2008; Sobel 2008), trade openness (Alcalá and Ciccone 2004; Frankel and Romer 1999; Melitz 2003), and population density (Combes et al. 2012; Glaeser et al. 1992).
This vector of control variables aims to account for key factors influencing the business environment beyond economic complexity. Higher tax burdens (contribution_rate) are expected to negatively affect firm entry and survival by reducing expected profitability and available resources for investment (Djankov et al. 2010; Tomasi et al. 2023). Minimum capital requirements (minimum_capital) represent a direct regulatory barrier to entry, potentially hindering entrepreneurship, especially for smaller ventures (Djankov et al. 2002; Sutton 1991; Tomasi et al. 2023). Conversely, a higher level of human capital (hc) within an economy signifies a more skilled workforce and a larger pool of potential entrepreneurs, which is expected to foster innovation, firm formation, and survival (Audretsch 2007; Chowdhury et al. 2019). Greater financial development (financial_development), proxied by domestic credit availability, is anticipated to alleviate financing constraints, thereby supporting both firm entry and the ability of existing firms to invest, grow, and survive shocks (Tomasi et al. 2023; Wang 2016). A higher degree of economic freedom (efw_score) generally indicates a more conducive environment for business, with lower regulatory burdens and stronger property rights potentially stimulating entry and firm performance (Bjørnskov and Foss 2008; Sobel 2008). The impact of trade openness (trade_openness) is theoretically ambiguous; while it can expand market opportunities and access to inputs, it also intensifies competitive pressures, with the net effect on firm dynamics often depending on firm productivity and sector characteristics (Alcalá and Ciccone 2004; Melitz 2003). Finally, population density (pop_density) serves as a proxy for agglomeration economies, which can offer benefits like larger markets and knowledge spillovers but also entail costs such as higher competition and congestion (Combes et al. 2012; Glaeser et al. 1992). Controlling these diverse factors allows for a more precise estimation of the specific impact of economic complexity on firm dynamics.
Table 1 presents the descriptive statistics of the variables. The descriptive analysis reveals considerable heterogeneity in the key variables employed in this study, spanning 24 European countries across various sectors and time periods. The Economic Complexity Index (ECI), the primary explanatory variable, exhibits a mean of 1.18, with a standard deviation of 0.45 and a substantial range (0.21 to 2.05), indicating significant variations in economic complexity across the country-sector observations. The firm dynamics variables, serving as the dependent variables, also display considerable dispersion. The average firm birth rate is 8.6%, while the death rate averages 8.4%, leading to a mean business churn rate of 17.5%. Survival rates, following the Eurostat definition (e.g., for 5 years [V97045]: "number of enterprises in the reference period (t) newly born in t-5 having survived to t divided by the number of enterprise births in t-5 - percentage"), decline over time as expected: on average, 85.4% of firms survive the first year, 68.5% survive to 3 years, and 56.4% survive to 5 years.
However, an observation emerges: although the official definition implies a theoretical maximum of 100%, the original descriptive statistics (see Table 11 at appendix) report extremely high and conceptually implausible maximum values for survival rates (e.g., 1200% for SR1, 1300% for SR3, 1700% for SR5). To address this, the top 5% of observations for each survival rate variable were removed from their respective distributions for the main econometric analysis, enhancing the robustness of our subsequent estimations. For transparency and completeness, descriptive statistics and regression analyses using the original, outlier-inclusive survival rate data are presented in Appendix A (e.g., Table 12). The control variables also vary considerably, reflecting the institutional and economic diversity within the sample: the average total tax and contribution rate is 42.8%, minimum capital requirements for starting a business vary widely (mean 16.4% of GDP per capita, std. dev. 17.5%), domestic credit averages 78.6% of GDP, and trade openness is high (mean 110.7% of GDP). The Human Capital Index and the Economic Freedom score show relatively less variation.
3.2 Estimation strategy
To investigate the impact of economic complexity (ECI) on firm dynamics, we begin by specifying a baseline empirical model. Our objective is to estimate the relationship between firm dynamics in a given country-sector i at year t and the level of economic complexity within the same country and year. Robust standard errors are also estimated, clustered at the country-sector level for firm dynamics. The model is adapted from prior work such as Tomasi et al. (2023) and Bottasso et al. (2017) and can be represented by the following equation:
where \(FirmDynamics_{icst}\) represents the different metrics of firm dynamics (birth rate, death rate, business churn, and 1-year, 3-year, and 5-year survival rates) for sector i, country c, firm-size class s, and year t; \(ECI_{ct}\) is the Economic Complexity Index for country c in year t; \(X_{ct}\) is a vector of relevant country-year control variables, including total tax and contribution rate, minimum capital requirement, human capital, domestic credit to the private sector, economic freedom, trade openness, and population density; \(\mu _{ic}\) represents country-sector fixed effects, which control for all time-invariant unobserved heterogeneity at this level, such as stable sectoral specialization patterns, institutional characteristics, regulatory environments, and geographic or historical advantages; and \(\varepsilon _{icst}\) is the idiosyncratic error term. The primary parameter of interest in this specification is \(\beta \), which captures the average association between country-level economic complexity and firm dynamics within country-sector cells over time.
However, our major hypothesis posits that the impact of ECI may vary with firm size. As developed in our theoretical framework (Section 2), the mechanisms through which ECI influences firm dynamics – both the ‘fertile ground’ opportunities and the ‘selective filter’ barriers – are unlikely to affect firms uniformly. A related example is provided by Tomasi et al. (2023) where the red tape cost is more pronounced in small firms. To test this heterogeneity, we extend the baseline model to include an interaction term between ECI and an indicator for firm size \(FirmSize_{cst}\), in this case, a dummy variable for microenterprises with fewer than five employees):
where \(FirmDynamics_{icst}\) denotes the outcome for country \(c\), sector \(i\), firm-size class \(s\), and year \(t\). \(ECI_{ct}\) is the country-year measure of economic complexity, \(SmallFirm_s\) is an indicator for micro-enterprises, and \(ECI_{ct} \times SmallFirm_s\) captures whether the relationship between complexity and firm dynamics differs for small firms. The term \(\mu _{ic}\) denotes country-sector fixed effects.
In this second specification, \(\beta _{1}\) captures the effect of ECI for the reference group (non-small firms), while the coefficient \(\beta _{2}\) captures the average difference between small and non-small firms when ECI equals zero. \(\beta _{3}\) measures the difference in this effect for the group of small firms. The total effect of ECI for small firms is therefore given by \(\beta _{1}~+ \beta _{3}\). The estimation of \(\beta _{3}\) is crucial for evaluating our hypothesis regarding the differential impact of complexity.
We perform several robustness checks. The first utilizes an instrumental variable (IV) approach estimated via two-stage least squares (2SLS) to address potential endogeneity between firm dynamics and economic complexity. This endogeneity may stem from two sources: (i) Potential omitted variables: The scope for omitted factors is influenced by elements such as free trade and the free movement of people within the European Union, which increase the possibility of regional spillovers through social pressure and imitation effects (Buera et al. 2011; Vu 2020), potentially affecting both local complexity and firm dynamics simultaneously. (ii) Reverse causality: Local firm dynamics – entry, survival, growth, and exit – may directly influence the observed level of complexity in that same country-year. This second source arises because an intrinsic characteristic of ECI is to reflect the productive structure (Hausmann and Hidalgo 2011; Hidalgo and Hausmann 2009), and firm dynamics are themselves a part of this same productive structure of the country.
To mitigate these two sources of endogeneity, we propose an instrument calculated as follows:
where \(N_{r}\) denotes the number of countries in region r. j stands for neighboring countries of domestic country i. For each country each year, we calculate a simple average of economic complexity of neighboring countries located in a particular region, excluding a country’s ECI values. This ‘leave-one-out regional average’ approach captures general trends in complexity development within a group of economically and/or geographically related countries (Gründler and Krieger 2016; Vu 2020). This instrumental variable strategy has been employed, for instance, by Acemoglu et al. (2019) to estimate causal effects on economic growth and specifically applied to the case of economic complexity by Vu (2020).
The rationale for using this instrument to control for endogeneity is twofold: (i) it captures broader trends in technological development, regional demand shocks, or knowledge diffusion that affect the entire region, including country A. These regional trends influence country A’s ECI (ensuring relevance) but are considered exogenous to the specific firm dynamics within country A; (ii) While country A’s ECI may be influenced by the firm dynamics within A, the average ECI of its regional neighbors (e.g., B and C) is not directly determined by the firm dynamics in A. The entry, exit, and growth decisions of firms in A have a negligible (or null) impact on the average ECI of its regional neighbors.
We also attempt to address the potential limitations of this instrument. Our identification argument follows the logic of regional-wave instruments, such as Acemoglu et al. (2019), in which regional variation is used to isolate diffusion-like changes that are not mechanically driven by the country’s own outcome. In our setting, the leave-one-out regional ECI captures common regional shifts in productive structures that are relevant for domestic ECI, while excluding the direct mechanical contribution of domestic firm dynamics to the country’s own ECI. Nevertheless, regional complexity may still be correlated with common regional shocks, trade integration patterns, or technological trajectories that also affect domestic firm dynamics. We therefore interpret the IV strategy as reducing, rather than fully eliminating, endogeneity concerns.
First, regarding the potential confounding effects of international trade, we include trade openness as a control variable to mitigate this impact. Second, concerning the choice of regions, we employ a specific regional classification for Europe, following Gründler and Krieger (2016) and Vu (2020) (See Appendix A). Furthermore, we report Anderson-Rubin (AR) confidence intervals for our main IV estimates. AR intervals are robust to the presence of weak instruments, providing valid inference regarding the uncertainty of the estimated effects, particularly in just identified models (Andrews et al. 2019; Huntington-Klein 2021; Keane and Neal 2023).
For all baseline and robustness checks, we estimate standard errors clustered at the country-sector level. We adopt this approach because it is highly probable that the remaining idiosyncratic errors \(\varepsilon _{it}\) within the same country-sector are correlated over time (serial correlation) (Abadie et al. 2023). Time-varying unobserved factors affecting firm dynamics (such as local sectoral shocks, gradual changes in uncaptured regulations, etc.) may persist over several years (Cameron and Miller 2015; MacKinnon et al. 2023).
4 Econometric findings
4.1 Fixed effects
Table 2 reports the baseline fixed-effects estimates relating economic complexity to firm dynamics. The estimates indicate a negative association between ECI and firm creation. A one-unit increase in ECI is associated with a 3.39 percentage point decline in the birth rate (\(p<0.01\)). To gauge the magnitude of this relationship, a one-standard-deviation increase in ECI (0.456) corresponds to a decline of about 1.55 percentage points, or roughly 18 percent of the sample mean birth rate. In economic terms, this is a sizeable association.
The reduction in entry is accompanied by lower overall turnover. A one-unit increase in ECI is associated with a 4.97 percentage point decline in business churn (\(p<0.01\)) and a 1.64 percentage point decline in the death rate (\(p<0.05\)). Taken together, these estimates suggest that more complex economies exhibit less firm turnover overall, with fewer firms entering and fewer firms exiting.
The estimates for survival are less precise. ECI is negatively associated with the 3-year survival rate, with an estimated effect of 7.23 percentage points (\(p<0.10\)). The coefficients for 1-year and 5-year survival are also negative, but they are not statistically distinguishable from zero in the fixed-effects specification. At this stage, the results point more clearly to lower entry and lower turnover than to a uniform effect of complexity on firm persistence. For this reason, the fixed-effects estimates are best interpreted as showing a strong negative relationship between complexity and firm creation, alongside more tentative evidence on medium- and long-run survival.
Table 3 examines whether these patterns differ by firm size. Small firms display higher birth, death, and churn rates, as well as lower survival rates at all horizons, which is consistent with the well-known liabilities of smallness. The interaction between ECI and the small-firm indicator shows that the relationship between complexity and firm dynamics is more pronounced for small firms, especially at entry.
For the birth rate, the estimated effect of ECI for non-small firms is -1.91 percentage points (\(p<0.10\)). The interaction term is -3.40 (\(p<0.01\)), implying a total marginal effect of -5.31 percentage points for small firms. Formally, the marginal effect for small firms is given by:
This pattern suggests that the negative relationship between complexity and entry is substantially stronger among small firms.
A similar pattern appears for firm exit and turnover. For the death rate, the baseline ECI coefficient for non-small firms is not statistically significant, whereas the interaction term is negative and statistically significant (-3.41). The implied marginal effect for small firms is -3.27 percentage points. For business churn, the baseline coefficient is again not statistically significant (-1.39), but the interaction term is large and negative (-7.22), implying a total effect of -8.62 percentage points for small firms. These estimates indicate that, among small firms, higher complexity is associated not only with lower entry but also with lower subsequent turnover.
The survival estimates are more nuanced. For non-small firms, ECI is associated with a 10.64 percentage point reduction in 3-year survival (\(p<0.01\)). The interaction term is positive (6.34), which implies a smaller negative effect of -4.30 percentage points for small firms. At the 5-year horizon, the baseline coefficient for non-small firms is negative but not statistically significant (-6.45), while the interaction term is positive and statistically significant (7.22), implying an estimated effect close to zero for small firms (0.78 percentage points).
Overall, the fixed-effects results point to three main patterns. First, higher economic complexity is associated with lower firm entry. Second, complexity is also associated with lower business turnover. Third, these relationships are more pronounced for small firms at the entry margin, while the negative association between complexity and longer-horizon survival appears weaker for small firms than for larger firms. These findings are consistent with the view that complexity is associated with stronger selection into markets, although the survival results remain more tentative and should be interpreted with caution until endogeneity is addressed more directly in the IV analysis.
4.2 Robustness checks: IV-2SLS
Table 4 reports the IV-2SLS estimates, which are intended to address endogeneity in the relationship between economic complexity and firm dynamics. The Kleibergen-Paap F-statistics range from 342 to 652, comfortably above conventional thresholds, indicating that the instrument is strong across specifications. The Anderson-Rubin confidence intervals also reinforce the main patterns by excluding zero for the estimated effects on birth rates, business churn, and medium-term survival.
Relative to the fixed-effects estimates, the IV results strengthen the negative association between ECI and firm creation. A one-unit increase in ECI is associated with a 5.51 percentage point decline in the birth rate (\(p<0.01\)), which is larger in magnitude than the corresponding fixed-effects estimate. This pattern is consistent with the view that endogeneity in the baseline specification may have attenuated the negative relationship between complexity and entry.
The IV results also point to a stronger relationship between complexity and firm persistence at longer horizons. ECI is associated with a 14.21 percentage point reduction in 3-year survival and a 10.73 percentage point reduction in 5-year survival, both statistically significant. By contrast, the coefficient on 1-year survival remains statistically indistinguishable from zero. Taken together, these estimates suggest that the effects of complexity on persistence are not concentrated in the very short run, but emerge more clearly over medium- and longer-term horizons.
For overall market dynamics, ECI is associated with a 5.10 percentage point decline in business churn, while the coefficient on the death rate remains statistically insignificant. This combination of lower entry, lower churn, and weaker evidence on firm exit per se suggests that complexity is more strongly related to reduced participation and reduced overall turnover than to a uniform increase in firm mortality.
Table 5 extends the IV analysis by incorporating heterogeneity by firm size. The baseline coefficients for non-small firms are generally more negative across several outcomes, while the interaction terms indicate that these negative relationships are attenuated for small firms in some dimensions.
The entry results continue to show that small firms are particularly exposed at the point of market entry. At the same time, the interaction estimates for medium- and long-term survival are positive, including SR3 (+8.81) and SR5 (+7.67), implying that the negative association between complexity and longer-horizon survival is weaker for small firms than for larger firms. The Anderson-Rubin confidence sets provide additional support for these heterogeneous effects.
Overall, the IV evidence reinforces three conclusions. First, higher economic complexity is associated with lower firm entry, and this relationship becomes stronger once endogeneity is addressed. Second, complexity is also associated with lower business turnover and lower survival at medium- and longer-term horizons. Third, the heterogeneous estimates indicate that the entry margin is especially restrictive for small firms, whereas the negative relationship between complexity and longer-run survival appears less severe for small firms than for larger firms. These results are consistent with a selective-filter interpretation of complexity, although the mechanisms behind the relative resilience of surviving small firms should be interpreted more cautiously and are better viewed as an open question for the discussion section.
4.3 Additional robustness: Permutation tests, sample sensitivity and wild bootstrap
To further assess the robustness of the main IV-2SLS findings, we perform several additional exercises. First, we implement permutation tests for the interaction between Economic Complexity (ECI) and the Small Firm indicator. Second, we conduct a leave-one-out country sensitivity analysis to verify whether the estimated interaction effects are disproportionately driven by any single country in the sample. Third, we reconstruct the instrumental variable using an alternative regional classification based on the United Nations Geoscheme.
We begin with permutation tests for the interaction term \(ECI \times Small\ Firm\), following the logic of randomization-based inference used in studies such as Chetty et al. (2009), Carrillo (2020), and Tian et al. (2024). Specifically, we shuffle ECI values across country-sector cells within each year and re-estimate the interaction specification 1,000 times. This procedure preserves the year-specific distribution of ECI while breaking the original assignment of ECI across country-sector observations. The resulting placebo estimates provide an empirical reference distribution under the null that the observed interaction pattern does not depend on the actual allocation of ECI in the data.
Figure 1 reports the results for the interaction term across the main firm-dynamics outcomes: Birth Rate, Death Rate, Business Churn, and the 1-, 3-, and 5-year Survival Rates. In all cases, the coefficients estimated from the original data lie in the extreme tails of the placebo distributions. The associated permutation p-values are very small, indicating that coefficients of similar magnitude would be unlikely to arise by chance under the proposed randomization scheme. These results provide additional support for the view that the heterogeneous effect of ECI by firm size is not an artifact of random assignment.
We next assess whether the interaction results are driven by any single country in the European sample. To do so, we re-estimate the IV-2SLS specification 24 times, excluding one country at a time and focusing on the \(ECI \times Small\ Firm\) coefficient for each firm-dynamics outcome.
Figures 2 through 7 in Appendix A show the resulting coefficients and 95% confidence intervals for Birth Rate, Death Rate, Business Churn, and the 1-, 3-, and 5-year Survival Rates. Overall, the estimated interaction effects remain stable in sign and broadly similar in magnitude across leave-one-out samples. Germany appears somewhat more influential for the 1- and 3-year survival outcomes, as its exclusion leads to somewhat larger changes in the estimated coefficient and wider confidence intervals than the exclusion of other countries. Even so, the qualitative pattern of results remains unchanged, indicating that the main heterogeneous effects are not driven by any single country.
We also reconstruct the leave-one-out instrument using the United Nations Geoscheme regional classification instead of the regional grouping proposed by Gründler and Krieger (2016), which is used in the baseline specification. Table 10 in Appendix A reports the results. The estimates remain directionally consistent with the main findings: higher economic complexity is associated with lower firm entry, and the interaction terms for 3- and 5-year survival remain positive for small firms. Although some coefficients are modestly smaller in magnitude, the central pattern is preserved. This suggests that the results are not overly sensitive to the particular regional classification used to construct the instrument.
As an additional conservative inference check, we report wild-cluster-bootstrap p values for the FE and IV-2SLS interaction models, clustering at the country level. This procedure is useful in our setting because the number of country clusters is limited and conventional cluster-robust inference may be sensitive to small-cluster asymptotics (Roodman et al. 2019). The results, reported in Table 13 in appendix A, support the presence of heterogeneous effects by firm size. In particular, the interaction between ECI and the small-firm indicator remains statistically significant or close to conventional significance levels for firm entry and business churn in both FE and IV-2SLS models. However, the marginal effects of ECI for small firms are less uniformly significant under this conservative inference. Therefore, we interpret the wild-bootstrap results as reinforcing the evidence of heterogeneous exposure by firm size, while also confirming that the strongest and most robust evidence concerns firm entry.
Finally, to facilitate the interpretation of the interaction models, Table 6 reports the marginal effects of ECI by firm size. For non-small firms, the marginal effect corresponds to the base coefficient on ECI. For small firms, the marginal effect is computed as the linear combination \(\beta _{ECI}+\beta _{ECI \times SmallFirm}\). The table reports standard errors, \(p\)-values, and confidence intervals for each marginal effect. Rather than introducing new evidence, this table is intended to consolidate the main heterogeneous patterns documented in the preceding regressions and robustness exercises.
5 Disentangling the effects of economic complexity on the firm dynamics
Our empirical analysis reveals a complex, multifaceted, and paradoxical relationship between economic complexity (ECI) and firm dynamics within the European context. The IV results in Tables 4 and 5 provide evidence that ECI functions simultaneously as an entry barrier, a stabilizing force for selected incumbents, and a differential survival determinant based on firm size. This duality refines the traditional ‘selective filter’ and ‘fertile ground’ mechanisms and shows that their effects are not symmetric across the firm size distribution.
5.1 The selective filter and entry barriers
Consistent with the selective filter hypothesis, higher ECI is associated with lower firm birth rates, with the negative effect substantially more pronounced for microenterprises. These firms face the compounded burden of sophisticated capability requirements, substantial sunk costs (Audretsch 1995; Sutton 1991), and dense tacit knowledge networks (Boschma 2005; Hartmann 2014), which erect formidable entry barriers. These constraints are amplified by the intrinsic “liability of smallness” (Cefis et al. 2022; Coad 2009) and limited capability vectors (Coad 2009; Storey 1994), restricting their ability to mobilize resources and interpret opportunity sets.
These barriers arise through multiple facets of complexity. Technological complexity requires absorptive capacity and R&D investment (Cohen and Levinthal 1990) that exceed most micro firms’ capabilities. Market complexity, characterized by intricate value chains, can obscure viable niches. Institutional complexity, consisting of regulatory and administrative environments, imposes fixed compliance costs that disproportionately burden small entrants (Audretsch 1995). Consequently, economic complexity acts as a capability-sensitive gatekeeper, filtering entrants according to their pre-existing alignment with capability thresholds established within the productive structure.
However, the survival results reveal a more nuanced pattern. The positive interaction term between ECI and small firm in the survival models suggests that economic complexity may attenuate the usual liability of smallness, conditional on entry. This does not imply that high-complexity environments are generally more favorable to small firms. Indeed, the entry results indicate that ECI reduces firm births more strongly among smaller firms. Rather, the survival results suggest a nonlinear selection pattern: complexity raises the entry threshold for small firms, but those that pass this threshold may be positively selected, better positioned in defensible niches, or more aligned with localized capability structures. In this sense, the relationship between complexity and small-firm dynamics is nonlinear: high ECI discourages entry, but among surviving firms it may reduce the survival disadvantage typically associated with small size.
5.2 Post-entry dynamics: Resilience and entrepreneurial conservatism
A more intricate dynamic unfolds for microenterprises that manage to pass this demanding entry barrier. Our IV results indicate that higher ECI is associated with lower initial death rates and lower churn among small firms. This suggests a powerful pre-selection effect: only the most capable, internally coherent, or strategically positioned microenterprises survive the entry filter. These firms may be able to effectively identify and exploit ‘fertile ground’ aspects of the ecosystem, such as localized knowledge spillovers (Audretsch 1995) or market interstices where competitive pressure is initially lower (Coad 2009). The heterogeneity within the SME population (Storey 1994) implies that these survivors are not typical micro firms but a selected subset with particular capabilities.
This resilience extends to longer-term persistence. While ECI is, on average, detrimental to long-term survival for all firms, its negative impact is markedly attenuated for microenterprises relative to larger firms. A compelling explanation advanced in this paper is the adoption of entrepreneurial conservatism. Surviving small firms may prioritize stability by operating in defensible niches, focusing on incremental improvements rather than aggressive scaling (Coad 2009; Dranove et al. 2016), and thereby insulating themselves from the higher-order competitive pressures associated with frontier innovation (Du and O’Connor 2021).
In contrast, larger firms are directly exposed to the full force of Schumpeterian competitive selection and innovation races (Aghion et al. 2015; Schumpeter 1942), which may impose escalating adaptation costs that over time degrade their survival probabilities. Thus, paradoxically, economic complexity may impose a heavier long-term adaptive burden on larger firms than on conservatively positioned microenterprises.
5.3 Lifecycle misallocation
These dynamics have broader implications for aggregate productivity. If the firms best adapted for persistence in high-ECI environments are conservative microenterprises rather than innovation-active “gazelles”, then the Schumpeterian mechanism of creative destruction may be distorted. We term this phenomenon lifecycle misallocation, where capital, labor, and managerial attention remain tied up in low-growth firms that survive due to strategic conservatism rather than productivity potential. At the same time, dynamic but risk-exposed ventures are filtered out at entry (Hsieh and Klenow 2009, 2014; Restuccia and Rogerson 2017).
These patterns are also likely to vary across sectors (Audretsch 1995). In knowledge-intensive business services (KIBS) and technology sectors, SMEs may leverage relatedness, proximity-based spillovers, and specialized niches (Caves 1998; Coad 2009; Sutton 1997) to participate in incremental innovation. By contrast, local, nontradable services primarily experience the cost-increasing aspects of complexity (Mishra et al. 2020), such as higher wages and rents, without enjoying compensatory capability-driven advantages. Such firms may adopt more conservative strategies (Coad et al. 2024), discouraging scaling and innovation.
The interpretation, however, requires caution. The observed resilience of small firms in high-complexity environments may reflect more than one mechanism. On the one hand, a survival-of-the-fittest process may operate at entry, whereby only the most capable or strategically aligned microenterprises are able to pass the initial complexity filter. On the other hand, surviving small firms may also persist through entrepreneurial conservatism, remaining in defensible niches and avoiding direct exposure to frontier competition. Since our data do not allow us to distinguish these mechanisms directly, we interpret the evidence as consistent with a combination of selective entry and conservative post-entry positioning.
In conclusion, these findings nuance established stylized facts about firm dynamics (Geroski 1995; Gil 2010; Manjón-Antolín and Arauzo-Carod 2008). In Europe, our evidence confirms that ECI acts as a powerful selective barrier at entry, particularly for smaller firms. Yet for those able to overcome this threshold, a pattern of relative post-entry resilience and comparatively lower long-run attrition emerges. This dynamic differs markedly from findings in lower-complexity or developing contexts, such as those reported by Ajide (2022), where economic complexity uniformly promotes entrepreneurship across the distribution rather than imposing capability-based filtering. This contrast underscores the context-contingent nature of complexity effects on firm dynamics and highlights the need for theoretical frameworks that explicitly integrate size, productive context, and institutional environment.
5.4 Policy recommendations
Our findings offer important policy considerations. Acknowledging the nature of economic complexity, it becomes clear that exploiting its potential requires a sophisticated policy design; simple, isolated instruments will not suffice. Given that market selection is a fundamental driver of productivity (Bartelsman et al. 2013; Foster et al. 2001), the policy objective should not be to eliminate the selective pressures of complex economies. Rather than defining an optimal or “fair” level of selection, the policy implication is more modest. In high-complexity environments, policy should reduce distortions that make firm selection depend excessively on financial constraints, administrative burdens, or unequal access to complementary capabilities, rather than on productivity, innovation potential, and capability development.
Firstly, facilitating access to growth finance is paramount. Operating in complex sectors often demands substantial investments in R&D and specialized capital (Sutton 1991), which can exacerbate the financial constraints typically faced by SMEs (Beck and Demirguc-Kunt 2006). An example is the ‘Valley of Death’, the structural funding and coordination gap between successful demonstration and market entry that frequently prevents technologies from advancing to commercialization (Beard et al. 2009; Frank et al. 1996). Therefore, policies should extend beyond basic start-up aid to encompass mechanisms such as government-backed loan guarantees and co-investment schemes, designed to actively facilitate the scale-up of efficient firms with growth potential.
Secondly, policies must promote the adoption of technology and digitalization. The demands of high-ECI environments require advanced absorptive and technological capabilities within firms (Cohen and Levinthal 1990). Policies aimed at strengthening digital infrastructures, interoperability standards, and data-sharing mechanisms therefore serve not only technical modernization purposes but also the systemic goal of enhancing learning and collective innovation. Active support through initiatives like subsidized technology extension services, dedicated digital transformation programs, and fostering collaborative R&D can help SMEs overcome these critical capability thresholds across heterogeneous firms and leverage technology for ambitious growth.
Thirdly, simplifying complex regulatory landscapes is necessary. High-ECI economies can impose disproportionately high administrative (red tape) costs on SMEs, with research showing that compliance time can be a greater impediment than direct monetary costs (Djankov et al. 2002; Tomasi et al. 2023). The goal should be to equalize the opportunity set, ensuring that firms of all sizes can compete and scale based on efficiency, not merely on resilience to red tape or structural advantages. Instead of merely simplifying rules or seeking perfect coordination, which is an unfeasible goal in complex systems characterized by uncertainty, regulation should be dedicated to supporting the emergence and co-evolution of new sectors and technologies (Pyka et al. 2023). By building an ecosystem that offsets the frictions induced by complexity, policy can transform the selective filter from an exclusionary force into a catalyst for dynamic and innovation-led growth. Policies in complex economies should therefore aim to reduce non-productivity-related barriers to entry and adaptation without weakening the role of market selection itself.
6 Conclusion
This study investigated the intricate relationship between productive structure, as captured by the Economic Complexity Index (ECI) at the country level, and the dynamics of firm entry, exit, and survival at the country-sector level in Europe, with a particular focus on how these effects vary by firm size. Our findings reveal a consistent and nuanced pattern. Higher economic complexity is robustly associated with lower firm entry rates and reduced long-term survival probabilities, especially after five years. Econometric results indicate that higher economic complexity significantly reduces firm entry and long-term survival, with more potent effects when accounting for endogeneity through instrumental variables. The relationship is marked by firm-size heterogeneity: microenterprises face steeper entry barriers but exhibit slightly better short-term survival, though their viability also declines over time under complex conditions. Taken together, the evidence supports the interpretation of economic complexity acting primarily as a selective filter in the European context. Complex productive environments appear to raise the threshold for entry and impose capability demands that erode firm survival over time. This filter is particularly stringent for firms lacking scale or adaptive capacity, although smaller firms that manage to enter may initially benefit from some resilience. These findings contribute to broader debates in the literature regarding the dual role of complexity in generating opportunities versus amplifying structural barriers.
This study is not without limitations. First, the analysis relies on aggregated country-sector data, which precludes a more granular investigation of intra-national disparities or the heterogeneity of firm behavior within sectors. In particular, the NACE two-digit classification used does not differentiate between capital- or technology-intensive industries and localized service sectors such as restaurants, potentially masking divergent mechanisms by which ECI affects different types of economic activity. Second, data quality remains a concern, particularly due to the presence of outliers in survival metrics, even though robust trimming and estimation techniques were employed. Third, the ECI is a proxy measure of productive capabilities and may not fully capture the multidimensional nature of complexity; alternative indicators or metrics could yield complementary insights. Finally, while the study discusses the importance of firm scalability, it does so only in the aggregate; it does not explore the varying growth potential across different firm types or business models.
Future research should build on these limitations by utilizing firm-level microdata to examine the precise mechanisms, such as innovation investment, financial constraints, or labor skill composition, through which complexity influences firm trajectories. In addition, more disaggregated sectoral studies distinguishing high-technology and knowledge-intensive industries from traditional or non-tradable services would provide clearer insight into the differentiated effects of complexity. Regional data could be helpful to analyze how the productive structure shapes firm dynamics in lagged regions in the European Union. Investigating heterogeneity in scalability potential, testing for nonlinear dynamics, and comparing institutional contexts across countries with different levels of economic development would also be promising directions. Despite these constraints, the study provides robust evidence that economic complexity, while potentially fertile, functions primarily as a gatekeeping mechanism in advanced economies, especially for smaller firms. Policies that help SMEs navigate this terrain are fundamental to ensuring that the benefits of economic complexity are broadly shared.
Data Availability
The datasets generated and analyzed during the current study are publicly available. Firm demographics data are available from Eurostat’s Business Demography Statistics (https://ec.europa.eu/eurostat Economic Complexity Index data are from the Observatory of Economic Complexity (https://oec.world/ Control variables data are from the World Bank, Penn World Table, and Fraser Institute. Replication code and additional materials are available from the corresponding author upon request.
Notes
Austria, Belgium, Bulgaria, Croatia, Czechia, Denmark, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Lithuania, Netherlands, Norway, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, Switzerland.
We use the sector aggregation from the database “Business demography by size class and NACE Rev. 2 activity (2004-2020)”.
References
Abadie A et al (2023) When should you adjust standard errors for clustering?*. In: The quarterly journal of economics 138(1), 1–35. ISSN: 0033-5533. https://doi.org/10.1093/qje/qjac038
Acemoglu D et al (2019) Democracy does cause growth. en. J Politic Econ 127(1):47–100. ISSN: 0022-3808, 1537-534X. https://doi.org/10.1086/700936 https://www.journals.uchicago.edu/doi/10.1086/700936
Acs Z, Audretsch DB (1990). Innovation and small firms. English. Cambridge, Mass.: MIT Press. ISBN: 978-0-262-01113-6
Aghion P, Akcigit U, Howitt P (2014) What do we learn from schumpeterian growth theory? en. In: Handbook of economic growth. vol. 2. Elsevier, pp 515–563. ISBN: 978-0-444-53546-7. https://doi.org/10.1016/B978-0-444-53540-5.00001-X https://linkinghub.elsevier.com/retrieve/pii/B978044453540500001X
Aghion P, Akcigit U, Howitt P (2015). The Schumpeterian Growth Paradigm. en. Ann Rev Econ 7(1):557–575. ISSN: 1941-1383, 1941-1391. https://doi.org/10.1146/annureveconomics-080614-115412 https://www.annualreviews.org/.doi/10.1146/annurev-economics-080614-115412
Ajide FM (2022) Economic complexity and entrepreneurship: insights from Africa. Int J Develop Issues 21(3):367–388. https://doi.org/10.1108/IJDI-03-2022-0047
Akcigit U, Ates ST (2023) What happened to US business dynamism? J Politic Econ 131(8):2059–2124. ISSN: 0022-3808. https://doi.org/10.1086/724289 https://www.journals.uchicago.edu/doi/abs/10.1086/724289
Alcalá F, Ciccone A (2004) Trade and productivity*. Quarter J Econ 119(2), 613–646. ISSN: 0033-5533. https://doi.org/10.1162/0033553041382139
Andrews I, Stock JH, Sun L (2019) Weak instruments in instrumental variables regression: Theory and practice. Ann Rev Econ 11(1):727–753. _eprint:. https://doi.org/10.1146/annurev-economics-080218-025643
Audretsch DB (1995) Innovation and industry evolution. en. Google-Books-ID: xbGbfSQWRMMC. MIT Press. ISBN: 978-0-262-01146-4
Audretsch DB (2007) Entrepreneurship capital and economic growth. Oxford Rev Econ Policy 23(1):63–78. https://doi.org/10.1093/oxrep/grm001
Audretsch DB, Feldman MP (1996) R&D spillovers and the geography of innovation and production. Am Econ Rev 86(3):630–640
Balland P-A et al (2020) Complex economic activities concentrate in large cities. en. Nat Human Behav 4(3):248–254. ISSN: 2397-3374. https://doi.org/10.1038/s41562-019-0803-3 https://www.nature.com/articles/s41562-019-0803-3
Balland P-A et al (2022) The new paradigm of economic complexity. Res Polic 51(3):104450. https://doi.org/10.1016/j.respol.2021.104450
Bartelsman E, Haltiwanger J, Scarpetta S (2013) Cross-country differences in productivity: The role of allocation and selection. Am Econ Rev 103(1):305–334. https://doi.org/10.1257/aer.103.1.305
Beard TR et al (2009) A valley of death in the innovation sequence: an economic investigation. Res Eval 18(5):343–356. ISSN: 0958-2029. https://doi.org/10.3152/095820209X481057
Beck T, Demirguc-Kunt A (2006) Small and medium-size enterprises: Access to finance as a growth constraint. J Bank Finan 30(11):2931–2943. ISSN: 0378-4266. https://doi.org/10.1016/j.jbankfin.2006.05.009 https://www.sciencedirect.com/science/article/pii/S0378426606000926
Bjørnskov C, Foss NJ (2008) Economic freedom and entrepreneurial activity: Some ˘across-country evidence. en. Public Choice 134(3):307–328. ISSN: 1573-7101. https://doi.org/10.1007/s11127-007-9229-y
Borsato A, Llerena P (2025) Inside the biotech: A dialogue between Rosenberg, Babbage, Smith, and Schumpeter. Econ Innov New Technol 0(0):1–33. ISSN: 1043-8599. https://doi.org/10.1080/10438599.2025.2526525
Borsato A, Llerena P (2026) The US university-industry link in the R&D of AI: Back to the origins? J Evolution Econ 36(1):17. ISSN: 1432-1386. https://doi.org/10.1007/s00191-025-00940-7
Boschma R (2005) Proximity and innovation: A critical assessment. Region Stud. https://doi.org/10.1080/0034340052000320887
Bottasso A, Conti M, Sulis G (2017) Firm dynamics and employment protection: Evidence from sectoral data. Labour Econ 48:35–53. ISSN: 0927-5371. https://doi.org/10.1016/j.labeco.2017.05.013 https://www.sciencedirect.com/science/article/pii/S0927537117302634
Buera FJ, Monge-Naranjo A, Primiceri GE (2011) Learning the wealth of nations. en. Econometric 79(1):1–45. _eprint: https://onlinelibrary.wiley.com/doi/pdf/10.3982/ECTA8299 ISSN: 1468-0262. https://doi.org/10.3982/ECTA8299 https://onlinelibrary.wiley.com/doi/abs/10.3982/ECTA8299
Bui AT et al (2021) Legal and financial constraints and firm growth: small and medium enterprises (SMEs) versus large enterprises. English. Heliyon 7(12). ISSN: 2405-8440. https://doi.org/10.1016/j.heliyon.2021.e08576 https://www.cell.com/heliyon/abstract/S2405-8440(21)02679-7
Caballero RJ (2010) Creative Destruction. Econ Growth. Palgrave Macmillan UK
Cameron AC, Miller DL (2015) A practitioners guide to clusterrobust inference. en. J Human Res 50(2):317–372. ISSN: 0022-166X, 1548-8004. https://doi.org/10.3368/jhr.50.2.317 https://jhr.uwpress.org/content/50/2/317
Carrillo B (2020) Present Bias and underinvestment in education? LongRun effects of childhood exposure to booms in colombia. J Labor Econ 38(4):1127–1265. ISSN: 0734-306X. https://doi.org/10.1086/706535 https://www.journals.uchicago.edu/doi/abs/10.1086/706535
Caves RE (1998) Industrial organization and new findings on the turnover and mobility of firms. J Econ Literat 36(4):1947–1982. ISSN: 0022-0515. https://www.jstor.org/stable/2565044
Cefis E et al (2022) Understanding firm exit: a systematic literature review. en. Small Bus Econ 59(2):423–446. ISSN: 1573-0913. https://doi.org/10.1007/s11187-021-00480-x
Chetty R, Looney A, Kroft K (2009) Salience and taxation: Theory and evidence. en. Am Econ Rev 99(4):1145–1177. ISSN: 0002-8282. https://doi.org/10.1257/aer.99.4.1145 https://www.aeaweb.org/articles?id=10.1257/aer.99.4.1145
Chowdhury F, Audretsch DB, Belitski M (2019) Institutions and entrepreneurship quality. Entrepreneur Theor Pract 43(1):51–81. https://doi.org/10.1177/1042258718780431
Ciccone A, Papaioannou E (2007) Red tape and delayed entry. J European Econ Assoc 5(2-3):444–458. ISSN: 1542-4766. https://doi.org/10.1162/jeea.2007.5.2-3.444
Coad A (2009) The growth of firms: A survey of theories and empirical evidence. en. Google-Books-ID: cpBFfrJXhzQC. Edward Elgar Publishing. ISBN:978-1-84844-910-7
Coad A et al (2024) Scale-ups and High-Growth Firms: Theory, Definitions, and Measurement. Springer Nature, Singapore. ISBN: 978-981-97-1378-3
Cohen WM, Levinthal DA (1990) Absorptive capacity: A new perspective on learning and innovation. Administrat Sci Quart 35(1):128152
Combes P-P et al (2012) The productivity advantages of large cities: Distinguishing agglomeration from firm selection. en. Econometrica 80(6):2543–2594. _eprint: https://onlinelibrary.wiley.com/doi/pdf/10.3982/ECTA8442 ISSN: 1468-0262. https://doi.org/10.3982/ECTA8442
Delgado M, Porter ME, Stern S (2010) Clusters and entrepreneurship. en. J Econ Geograp 10(4):495–518. ISSN: 1468-2702, 1468-2710. https://doi.org/10.1093/jeg/lbq010 https://academic.oup.com/joeg/article-lookup/doi/10.1093/jeg/lbq010
Djankov S et al (2002) The regulation of entry*. Quart J Econ 117(1):1–37. ISSN: 0033-5533. https://doi.org/10.1162/003355302753399436
Djankov S et al (2010) The effect of corporate taxes on investment and entrepreneurship. en. Am Econ J: Macroecon 2(3):31–64. ISSN: 1945-7707. https://doi.org/10.1257/mac.2.3.31 https://www.aeaweb.org/articles?id=10.1257/mac.2.3.31
Dosi G (2023) The Foundations of Complex Evolving Economies: Part One: Innovation, Organization, and Industrial Dynamics. Oxford University Press, Oxford, New York. ISBN: 978-0-19-286604-2
Dranove D et al (2016) Economics of Strategy. Wiley, Hoboken, NJ. ISBN: 978-1-119-04231-0
Du K, O’Connor A (2021) Examining economic complexity as a holistic innovation system effect. en. Small Bus Econ 56(1):237–257. ISSN: 1573-0913. https://doi.org/10.1007/s11187-019-00215-z
Duranton G, Puga D (2004) Chapter 48 - Micro-foundations of urban agglomeration economies. Handbook of regional and urban economics. Elsevier
Ericson R, Pakes A (1995) Markov-perfect industry dynamics: A framework for empirical work. en. Rev Econ Stud 62(1):53. ISSN: 00346527. https://doi.org/10.2307/2297841 https://academic.oup.com/restud/article-lookup/doi/10.2307/2297841
Foster L, Haltiwanger JC, Krizan CJ (2001) Aggregate productivity growth: Lessons from microeconomic evidence. New Developments in Productivity Analysis. University of Chicago Press
Frank C et al (1996) Surviving the valley of death: A comparative analysis. en. J Technol Transfer 21(1):61–69. ISSN: 1573-7047. https://doi.org/10.1007/BF02220308
Frankel JA, Romer DH (1999) Does trade cause growth? Am Econ Rev 89(3):379–399. https://doi.org/10.1257/aer.89.3.379
Freitas E, Britto G, Amaral P (2024) Related industries, economic complexity, and regional diversification: An application for Brazilian microregions. Papers Region Sci 103(1):100011. ISSN: 1056-8190. https://doi.org/10.1016/j.pirs.2024.100011 https://www.sciencedirect.com/science/article/pii/S1056819024000290
Frenken K, Van Oort F, Verburg T (2007) Related variety, unrelated variety and regional economic growth. Region Stud 41(5):685–697. https://doi.org/10.1080/00343400601120296
Geroski PA (1995) What do we know about entry? Intern J Indust Organ. Post-Entry Perf Firms 13(4):421–440. ISSN: 0167-7187. https://doi.org/10.1016/0167-7187(95)00498-X https://www.sciencedirect.com/science/article/pii/016771879500498X
Gil PM (2010) Stylised facts and other empirical evidence on firm dynamics, business cycle and growth. Res Econ 64(2):73–80. ISSN: 1090-9443. https://doi.org/10.1016/j.rie.2009.11.001 https://www.sciencedirect.com/science/article/pii/S1090944309000568
Glaeser EL et al (1992) Growth in Cities. J Politic Econo 100(6):1126–1152. https://doi.org/10.1086/261856
Gründler K, Krieger T (2016) Democracy and growth: Evidence from a machine learning˘aindicator. European J Politic Econ. Institut Well Being 45:85–107. ISSN: 0176-2680. https://doi.org/10.1016/j.ejpoleco.2016.05.005 https://www.sciencedirect.com/science/article/pii/S0176268016300222
Haltiwanger JC, Bartelsman E, Scarpetta S (2004) Microeconomic evidence of creative destruction in industrial and developing countries. The World Bank
Hartmann D (2014) Economic complexity and human development: How economic diversification and social networks affect human agency and welfare. Taylor & Francis
Hausmann R, Hidalgo CA (2011) The network structure of economic output. J Econ Growth 16(4):309–342. https://doi.org/10.1007/s10887-011-9071-4
Hausmann R, CA Hidalgo CA et al (2014) The atlas of economic complexity: Mapping paths to prosperity. Mit Press
Hidalgo C (2015) Why Information Grows: The Evolution of Order, from Atoms to Economies. Basic Books, New York. ISBN: 978-0-465-04899-1
Hidalgo CA, Stojkoski V (2025) The theory of economic complexity. https://doi.org/10.48550/arXiv.2506.18829 arXiv: 2506.18829 [econ]
Hidalgo C et al (2007) The product space conditions the development of nations. Science 317(5837):482–487. https://doi.org/10.1126/science.1144581
Hidalgo C, Hausmann R (2009) The building blocks of economic complexity. Proceed Nat Acad Sci 106(26):1057010575
Hopenhayn H (1992) Entry, exit, and firm dynamics in long run equilibrium. Econometrica 60(5):1127–1150. ISSN: 0012-9682. https://doi.org/10.2307/2951541 https://www.jstor.org/stable/2951541
Hsieh C-T, Klenow PJ (2014) The life cycle of plants in India and Mexico *. Quarterl J Econ 129(3):1035–1084. ISSN: 0033-5533. https://doi.org/10.1093/qje/qju014
Hsieh C-T, Klenow PJ (2009) Misallocation and manufacturing TFP in China and India. Quarterl J Econ 124(4):1403–1448. https://doi.org/10.1162/qjec.2009.124.4.1403
Huntington-Klein N (2021) Instrumental variables. The Effect. Num Pages: 36. Chapman and Hall/CRC. ISBN: 978-1-003-22605-5
Jovanovic B (1982) Selection and the evolution of industry. Econometrica 50(3):649–670. ISSN: 0012-9682. https://doi.org/10.2307/1912606 https://www.jstor.org/stable/1912606
Keane M, Neal T (2023) Instrument strength in IV estimation and inference: A guide to theory and practice. J Economet 235(2):1625–1653. ISSN: 0304-4076. https://doi.org/10.1016/j.jeconom.2022.12.009 https://www.sciencedirect.com/science/article/pii/S0304407623000222
Klepper S (1996) Entry, exit, growth, and innovation over the product life cycle. Am Econ Rev 86(3):562–583
Koch P (2021) Economic complexity and growth: Can value-added exports better explain the link? Econ Lett 198:109682. https://doi.org/10.1016/j.econlet.2020.109682
Lundvall B (2010) National systems of innovation: Toward a Theory of Innovation and Interactive Learning. Anthem Press. ISBN: 978-0-85728-674-1
MacKinnon JG, Nielsen MØ, Webb MD (2023) Cluster-robust inference: A guide to empirical practice. J Economet 232(2):272–299. ISSN: 03044076. https://doi.org/10.1016/j.jeconom.2022.04.001
Malerba F (2004) Sectoral systems of innovation: Concepts, issues and analyses of six major sectors in Europe. Cambridge University Press. ISBN: 978-1-139-45416-2
Manjón-Antolín MC, Arauzo-Carod J-M (2008) Firm survival: methods and evidence. en. Empirica 35(1):1–24. ISSN: 1573-6911. https://doi.org/10.1007/s10663-007-9048-x
Marshall A (1890) Principles of economics Macmillan. In: London (8th ed.Published in 1920)
Mealy P, Doyne Farmer J, Teytelboym A (2019) Interpreting economic complexity. Sci Adv 5(1):eaau1705. ISSN: 2375-2548. https://doi.org/10.1126/sciadv.aau1705
Melitz MJ (2003) The impact of trade on intra-industry reallocations and aggregate industry productivity. Econometrica 71(6):1695–1725
Mishra S, Tewari I, Toosi S (2020) Economic complexity and the globalization of services. Struct Chang Econ Dyn 53:267280
Moretti E (2004) Workers’ education, spillovers, and productivity: evidence from plant-level production functions. Am Econ Rev 94(3):656690
Neffke F, Martin H, Boschma R (2011) How do regions diversify over time? Industry relatedness and the development of new growth paths in regions. Econ Geograp 87(3):237–265. _eprint: https://www.tandfonline.com/doi/pdf/10.1111/j.1944-8287.2011.01121.x ISSN: 0013-0095. https://doi.org/10.1111/j.1944-8287.2011.01121.x https://www.tandfonline.com/doi/abs/10.1111/j.1944-8287.2011.01121.x
Nelson RR, Winter SG (1982) An evolutionary theory of economic change. en. Google-Books-ID: 6Kx7s_HXxrkC. Harvard University Press. ISBN: 978-0-674-04143-1
Nightingale P, Coad A (2014) Muppets and gazelles: political and methodological biases in entrepreneurship research. Indust Corporate Change 23(1):113–143. ISSN: 0960-6491. https://doi.org/10.1093/icc/dtt057
North DC (1990) Institutions, Institutional Change and Economic Performance. Cambridge University Press, Cambridge. ISBN: 978-0-521-39416-1
OECD and Statistical Office of the European Communities (2008) Eurostat-OECD manual on business demography statistics. OECD. ISBN: 978-92-64-04187-5
OECD (2021) Understanding Firm Growth: Helping SMEs Scale Up. Organisation for Economic Co-operation and Development, Paris
Pyka A, Ari E, Lang S (2023) 18: Dedicated regulation: translating missions into regulation. en. In: Handbook of Innovation and Regulation. Section: Handbook of Innovation and Regulation. ISBN: 978-1-80088-447-2. URL: https://www.elgaronline.com/edcollchap/book/9781800884472/b-9781800884472.00027.xml
Rajan RG, Zingales L (1998) Financial dependence and growth. Am Econ Rev 88(3):559–586
Restuccia D, Rogerson R (2017) The causes and costs of misallocation. J Econ Perspect 31(3):151–174. https://doi.org/10.1257/jep.31.3.151
Roodman D et al (2019) Fast and Wild: Bootstrap inference in stata using boottest. Stata J 19(1):4–60. ISSN: 1536-867X. https://doi.org/10.1177/1536867X19830877
Schumpeter JA (1934) The theory of economic development: an inquiry into profits, capital, credit, interest, and the business cycle. Harvard University Press
Schumpeter JA (1942) Capitalism, Socialism and Democracy. Harper & Row, New York
Sobel RS (2008) Testing Baumol: Institutional quality and the productivity of entrepreneurship. J Bus Ventur. Econ Entrepreneur 23(6):641–655. ISSN: 0883-9026. https://doi.org/10.1016/j.jbusvent.2008.01.004 https://www.sciencedirect.com/science/article/pii/S0883902608000086
Stojkoski V, Koch P, Hidalgo CA (2023) Multidimensional economic complexity and inclusive green growth. Commun Earth Environm 4(1):1–12. https://doi.org/10.1038/s43247-023-00770-0
Storey DJ (1994) Understanding The Small Business Sector. Routledge, London. ISBN: 978-1-315-54433-5
Sutton J (1991) Sunk costs and market structure: Price competition, advertising, and the evolution of concentration. en. Google-Books-ID: HfxcCO5msgwC. MIT Press. ISBN: 978-0-262-19305-4
Sutton J (1997) Gibrat’s Legacy. J Econ Literat 35(1):40–59. ISSN: 0022-0515. https://www.jstor.org/stable/2729692
Tian L et al (2024) Digital finance and corporate leverage manipulation: Evidence from China. en. J Knowl Econ. ISSN: 1868-7873. https://doi.org/10.1007/s13132-024-02187-2
Tomasi C, Pieri F, Cecco V (2023) Red tape and industry dynamics: a cross-country analysis. J Ind Bus Econ 50(2)283 320. https://doi.org/10.1007/s40812-023-00266-0
Vu TV (2020) Economic complexity and health outcomes: A global perspective. Soc Sci Med 265:113480. ISSN: 0277-9536. https://doi.org/10.1016/j.socscimed.2020.113480 https://www.sciencedirect.com/science/article/pii/S0277953620306997
Wang Y (2016) What are the biggest obstacles to growth of SMEs in developing countries? An empirical evidence from an enterprise survey. Borsa Istanbul Rev 16(3):167–176. https://doi.org/10.1016/j.bir.2016.06.001
Acknowledgements
The author Alan Bueno thanks the institutional support of the University of Hohenheim where part of this article was written. All authors are grateful for the suggestions given in the seminars of the Center for Studies on Productive Structure, Human Development, and Sustainability (NEPDHS) and the University of Hohenheim seminars. We sincerely thank the handling editor and the two anonymous reviewers for their helpful comments and suggestions, which substantially improved the manuscript. All remaining errors are our own.
Funding
The Article Processing Charge (APC) for the publication of this research was funded by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) (ROR identifier: 00x0ma614). This work was supported by the National Council for Scientific and Technological Development (CNPq) [grant numbers 311036/2022-8, 406726/2025-6, and 88881.218631/2025-01] and the Brazilian Federal Agency for Support and Evaluation of Graduate Education (CAPES) [PhD Scholarship, Code 0001].
Author information
Authors and Affiliations
Contributions
Conceptualization, Methodology, Formal analysis, Data curation, Software, Writing - original draft: AB. Review & editing: AP. Supervision, Funding acquisition, Writing - review & editing: DF.
Corresponding author
Ethics declarations
Competing Interests
Andreas Pyka serves as an Advisory Editor of the Journal of Evolutionary Economics. He had no role in the editorial handling of this submission. The authors declare no other competing interests.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Below is the link to the electronic supplementary material.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/
About this article
Cite this article
Bueno, A., Pyka, A. & Ferraz, D. Economic complexity and firm dynamics: Evidence from European industries. J Evol Econ 36, 62 (2026). https://doi.org/10.1007/s00191-026-00980-7
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1007/s00191-026-00980-7