Beyond Firm Size: Managerial Practices, Sustainability Upgrading and Export Readiness in Ecuadorian Firms

Abstract

Export participation in emerging economies depends not only on structural firm characteristics but also on internal organizational capabilities that support international readiness. This study examines whether managerial practices and sustainability are associated with export participation among Ecuadorian firms and whether this relationship coexists with complementary capabilities such as quality certification and process innovation. Using firm-level data from the World Bank Enterprise Survey Ecuador 2024, the study applies a quantitative, non-experimental, cross-sectional design and estimates weighted logit models with region and sector fixed effects, complemented by robustness checks including alternative specifications, penalized logit, alternative inference for few clusters, and models of export intensity. The results show that managerial practices are the most consistent predictor of export participation in the main specifications. Sustainability presents a weaker but positive association, particularly when measured as the intensity of sustainable practices rather than as a binary condition. Quality certification emerges as the most consistent complementary capability, while process innovation shows a positive but less stable relationship. Additional analyses indicate that these capabilities are more clearly associated with the decision to export than with export intensity. The findings suggest that export readiness in Ecuadorian firms is linked to management, standardization, and gradual sustainability upgrading.

1. Introduction

Growing competitiveness in domestic and international markets has intensified interest in identifying the internal capabilities that enable firms to establish a stronger foothold in foreign trade. In emerging economies, this issue is particularly salient because many firms face constraints relating to scale, finance, access to standards, technological adoption, and organizational capabilities. In this context, the literature has shown that the quality of business management is not merely an administrative attribute, but an organizational capability that shapes firm performance, adaptability, and competitiveness (Bloom & Van Reenen, 2007; Quinn & Woodruff, 2021; Scur et al., 2021; Valero, 2021).

In Ecuador, this discussion is of particular importance. The business structure combines a broad base of firms oriented towards the domestic market with a relatively small group engaged in exports. The World Bank Enterprise Survey 2024 makes it possible to examine this issue using comparable data on managerial practices, innovation, certification, sustainability, and firm performance in a representative sample of the formal private sector; however, evidence for Ecuador on the link between management, sustainability, and exporting remains limited, particularly from an approach that integrates these dimensions within a single empirical framework (World Bank, 2024, 2025a).

Recent literature suggests that export participation does not depend solely on structural attributes such as size, age, or foreign ownership, but also on internal capabilities of organization, coordination, standardization, and continuous improvement. From this perspective, managerial practices may operate as a strategic capability for export readiness, while sustainability, certification, and process innovation act as complementary dimensions of that readiness (Bloom & Van Reenen, 2007; Cirera & Maloney, 2017; Quinn & Woodruff, 2021; World Bank, 2025c).

Although the literature has examined managerial practices, sustainability, and exporting separately, there is limited evidence assessing their joint articulation in firms from emerging economies facing organizational capability constraints. Much of the available evidence continues to focus on firms’ structural attributes, while the combined role of internal organizational capabilities has received less attention, especially in contexts where constraints on learning, standardization, and upgrading are significant.

Ecuador constitutes a theoretically informative case because it combines a business base largely oriented towards the domestic market with a relatively small share of exporting firms, in an environment where international integration coexists with scale constraints, heterogeneous organizational capabilities, and uneven adoption of standards and sustainable practices. In this sense, the Ecuadorian case makes it possible to observe whether internal capabilities such as management, certification, and sustainability are associated with export participation even in a context where structural attributes do not fully explain firm heterogeneity.

Accordingly, this study examines whether managerial practices and sustainability are associated with the export participation of Ecuadorian firms and whether this relationship coexists with complementary capabilities such as quality certification and process innovation. The article makes a twofold contribution: first, it provides recent micro-level evidence for Ecuador; second, it uses export readiness as an integrative lens for relating management, sustainability, standardization, and operational innovation into a single empirical framework. Here, export readiness refers to the organizational preparedness to meet the coordination, standardization, adaptation, and upgrading demands associated with foreign-market participation; it is not measured as a separate latent construct or interpreted as a causal mechanism.

The results show that managerial practices are positively associated with the probability of exporting in the principal specifications, whereas the evidence for sustainability is weaker and varies with its measurement. Quality certification displays the clearest association among the complementary capabilities. The remainder of the article is organized as follows. Section 2 reviews the literature and sets out the conceptual framework. Section 3 describes the data source, variables, and econometric strategy. Section 4 presents the results. Section 5 discusses the implications of the study, and Section 6 concludes.

2. Literature Review and Conceptual Framework

2.1. Management and Firm Heterogeneity

The management literature has shown that managerial practices constitute an important source of heterogeneity in firm performance. Bloom and Van Reenen argue that differences in monitoring, target-setting, and incentives are associated with productivity, profitability, growth, and firm survival. Building on this contribution, the literature has consolidated the view that management should not be understood merely as an administrative activity, but as an organizational capability that can be measured and compared across firms and countries (Bloom & Van Reenen, 2007; Quinn & Woodruff, 2021; Scur et al., 2021). This perspective is particularly relevant for emerging economies, where constraints arise not only from limited access to finance or technology, but also from deficits in internal organization that restrict knowledge absorption, process improvement, and productive upgrading (Cirera & Maloney, 2017; Cirera et al., 2020). From this standpoint, managerial practices may be understood as a foundational capability that conditions the possibility of activating more complex organizational mechanisms.

2.2. Exporting and Organizational Capabilities

From the perspective of firm-level international trade, exporting requires more than producing efficiently. It entails compliance with standards, operational coordination, consistent quality, responsiveness to external requirements, and the management of relationships with buyers and value chains. This logic is consistent with the view that exporting depends not only on observable endowments, but also on internal capabilities for coordination and adaptation (Valero, 2021; Feng & Valero, 2020).

Within this framework, managerial practices do not replace the classical determinants of exporting; rather, they may operate as an organizational capability that facilitates the activation of other complementary mechanisms, such as certification, operational innovation, or the adoption of standardized routines. In this way, export heterogeneity may be understood as the result of a combination of structural attributes and internal capabilities.

2.3. Sustainability, Standards, and Upgrading

The relationship between sustainability and export performance has gained prominence in the recent literature. Several studies show that green export strategies and ESG performance are positively associated with international competitiveness, although their effects are often mediated by export readiness, green innovation, and reduced financial constraints (Singh et al., 2024; Cai & Hao, 2025).

This nuance is important. Sustainability may enhance reputation, access to buyers, regulatory compliance, and efficiency in resource use, but its export-related benefits do not necessarily operate automatically or uniformly. The literature on international standards and certification complements this argument: the adoption of standards and certifications has become increasingly relevant for market functioning, technological diffusion, and export growth; at the firm level, certification may be interpreted as a signal of deeper internal capabilities (World Bank, 2025c, 2025b; Yang et al., 2023; Hunady & Chyláková, 2024).

In emerging economies, these complementary capabilities are especially decisive because they help close the gap between export intention and the effective capacity to compete in foreign markets. The literature on innovation and trade has shown that process innovation is relevant to exporting in developing economies: it supports SME internationalization, improves operational efficiency, and reinforces the production consistency required to compete internationally (Cassiman & Golovko, 2011; Edeh et al., 2020; Van Beveren & Vandenbussche, 2010; Ayllón & Radicic, 2019).

2.4. Alternative Explanations and the Study’s Position

The conceptual framework adopted in this study does not assume fully identified causal relationships between management, sustainability, and exporting. In a cross-sectional design, managerial practices may operate as an antecedent, a correlate, or even a partial outcome of processes of international integration. Similarly, quality certification may be interpreted as a prior capability that facilitates exporting, but also as an adaptive response once the firm is already participating in foreign markets.

Accordingly, the article does not identify causal mediation or strict temporal sequences, but rather conditional associations consistent with the view that export participation coexists with a set of internal organizational capabilities. From this position, the contribution of the study lies in integrating these dimensions within a single empirical framework and assessing whether they display patterns consistent with a logic of export readiness in the context of organizational constraints.

2.5. Hypotheses

This article adopts an export-readiness perspective in which managerial practices, sustainability intensity, quality certification, and process innovation represent differentiated dimensions of organizational preparedness for internationalization. Managerial practices capture internal coordination, sustainability intensity reflects cumulative upgrading, quality certification represents verifiable standardization, and process innovation captures operational adaptation. Considering these dimensions together shifts attention from isolated predictors toward the organizational conditions associated with international engagement. The framework does not imply a causal sequence or assume that these capabilities operate with equal empirical strength (Bloom & Van Reenen, 2007; Cassiman & Golovko, 2011; Yang et al., 2023; Singh et al., 2024; Cai & Hao, 2025; World Bank, 2025c).

H1. 

Managerial practices are positively associated with firms’ export participation.

H2. 

The intensity of sustainable adoption is positively associated with export participation.

H3. 

Quality certification is positively associated with export participation.

H4. 

Process innovation is positively associated with export participation and may function as a complementary capability within the relationship between management and exporting.

Given the cross-sectional design of the study, these hypotheses are formulated in terms of conditional associations consistent with the conceptual framework, rather than as fully identified causal relationships.

3. Methodology

3.1. Research Design and Data Source

This study adopts a quantitative, non-experimental, cross-sectional design based on the World Bank Enterprise Survey (WBES) Ecuador 2024. The WBES is a firm-level survey administered to a representative sample of the formal private sector and collects information on firm characteristics, innovation, organizational practices, certification, sustainability, and international integration (World Bank, 2024, 2025c).

The empirical strategy uses different analytical subsamples according to data availability and the dependent variable in each model. As shown in

Table 1. Sample and main descriptive statistics.

Panel (A) Analytical samples
Sample	Main Criterion			N	Strata
df_green	Management and sustainability			173	33
df_cert	Management and certification			171	33
df_proc	Management and process innovation			146	31
df_export	Exports, sustainability and mechanisms			144	31
Panel (B) Descriptive statistics for the export sample
Variable	Mean	Std. Dev.	Median	Min.	Max.
mp_core	0.587	0.161	0.613	0.083	0.926
green_index	0.531	0.398	0.500	0.000	1.000
ln_emp	4.506	1.301	4.283	1.792	7.550
age	31.528	21.571	25.000	3.000	118.000
foreign_share	11.333	28.586	0.000	0.000	100.000
export_share	7.326	19.866	0.000	0.000	99.000
Panel (C) Indicators for the export sample
Indicator	Unweighted			Weighted
Probability of exporting (export_dummy)	0.292			0.229
Sustainable adoption (green_adopt)	0.715			0.610
Quality certification (quality_cert)	0.438			0.305
Process innovation (proc_innov)	0.556			0.508
Mean mp_core	0.587			0.579
Panel (D) Weighted probability of exporting by organizational capabilities
Variable	Group	Weighted Probability of Exporting		N
green_adopt	0	0.117		41
green_adopt	1	0.301		103
quality_cert	0	0.124		81
quality_cert	1	0.469		63
proc_innov	0	0.197		64
proc_innov	1	0.261		80

Notes: calculations based on the export sample (n = 144).

, Panel A, the cleaned dataset contains 175 observations and 46 variables. The article’s main sample corresponds to the export equation, with 144 observations and 31 strata; in the logit models with absorbed fixed effects, the effective sample is reduced to 129 observations owing to the elimination of groups with no variation in the dependent variable. This decision preserves internal comparability across specifications and avoids artificial changes in sample composition arising from missing values.

3.2. Study Variables

The main dependent variable is export_dummy, which takes the value 1 if the firm engages in exporting and 0 otherwise. This variable approximates export participation, understood as an observable manifestation of firm-level international integration. The central explanatory variable is mp_core, a continuous index of managerial practices. Sustainability is incorporated in two ways: First, through green_adopt, a binary variable capturing sustainable adoption. Second, through green_index, a continuous index intended to capture the intensity of sustainable practices. Complementary organizational capabilities include quality_cert, which identifies the presence of quality certification, and proc_innov, which captures process innovation. The vector of controls includes ln_emp, age, and foreign_share, variables commonly used in the literature on firm performance and export participation.

3.3. Index Construction and Sample Treatment

The indices used in the article summarize observable organizational dimensions of firms. mp_core is defined as the firm-level simple average of nine core management items (r1, r2, r4, r5, r6, r7, r8, r10, and r11), recoded according to the transformations reported in Supplementary Table S1 and combined into a composite index. The index summarises the coded response profile across the selected management dimensions and is interpreted as an empirical composite measure rather than as a strictly monotonic ranking of managerial quality across every response category. Formally, $mp\_cor{e}_i={\displaystyle \frac{1}{n_i}}{\sum }_{j\in J}{\tilde{r}}_{ij}$, where J = {r1, r2, r4, r5, r6, r7, r8, r10, r11} and n_i corresponds to the number of non-missing responses in that set; the variable mp_core_n records this count, and mp_core was retained when at least six components were observed. The green_index is constructed as the simple average of two binary indicators of observed sustainable practices (ge7_yes and ge8d_yes), and therefore also ranges between 0 and 1. The variable green_adopt takes the value 1 when the firm reports at least one of these practices, 0 when at least one component is observed and neither practice is reported, and is missing when both components are missing. The nine management components are identified by their original WBES variable codes to ensure reproducibility. Their exact questionnaire wording, response categories, recoding rules, and the number of valid observations used for each item are reported in Supplementary Table S1, based directly on the Ecuador 2024 questionnaire and the implemented coding definitions. Items r3 and r9 were excluded from mp_core because they are conditional follow-up questions. Item r3 is answered only by firms that report monitoring performance indicators in r2, whereas r9 is answered only by firms that report managerial performance bonuses in r8. Including these items would make their availability dependent on questionnaire skip patterns and would assign additional weight to the monitoring and incentive dimensions already represented by r2 and r8.

The sample selection strategy is based on analytical subsamples constructed according to the simultaneous availability of information for the dependent variable and the relevant regressors in each model. The cleaned dataset contains 175 observations and 46 variables, whereas the main export sample comprises 144 observations and 31 strata. In the logit models with region and sector fixed effects, the effective sample is reduced to 129 observations owing to the elimination of one sector with no variation in the dependent variable. This reduction does not reflect a new substantive data-cleaning step, but rather the identification logic of fixed-effects models.

In terms of missing-data treatment, each subsample was constructed by excluding observations with missing values in the variables required for the corresponding estimation. This criterion preserves internal comparability within each block of models, although it implies minor differences in sample size between auxiliary and main equations.

3.4. Econometric Strategy

The empirical strategy is organized into three blocks. First, auxiliary equations are estimated to assess whether managerial practices are associated with sustainability, quality certification, and process innovation. Second, export models are estimated in which export participation is related to managerial practices and sustainability. Third, quality certification and process innovation are incorporated to examine whether the association between management and exporting is attenuated once these complementary capabilities are taken into account.

The main model is represented schematically as:

logit(P(export_dummy_i = 1)) = β₀ + β₁ mp_core_i + β₂ Green_i + γ′X_i + δ_s + τ_r,

where Green_i alternatively denotes green_adopt or green_index; X_i is the vector of controls; and δ_s and τ_r correspond to sector and region fixed effects, respectively. An extended version is then estimated incorporating quality_cert_i and proc_innov_i as complementary organizational capabilities.

3.5. Estimation Method and Robustness

The main estimation method is a weighted logit model with absorbed region and sector fixed effects, implemented using feglm() from the fixest package. This choice is justified because the main dependent variable is binary, the survey has a stratified design, and the absorbed fixed effects make it possible to control for territorial and sectoral heterogeneity without unnecessarily inflating the number of parameters. All main models use wmedian as the baseline weight and robust standard errors clustered by stratum. The baseline specification uses wmedian because it corresponds to the WBES median eligibility assumption, while wweak and wstrict are used exclusively as sensitivity checks. In the Ecuador 2024 survey, the strict assumption includes eligibility codes 1, 2, 3, 4, and 16; the median assumption additionally includes codes 10, 11, and 13; and the weak assumption additionally includes codes 91, 92, 93, 94, and 12. These weights were used as supplied in the WBES dataset and were not trimmed, recalibrated, or reconstructed by the authors. Absorbing region and sector fixed effects controls for common location- and industry-specific heterogeneity without estimating and displaying a large set of dummy coefficients. Nevertheless, groups with no within-group variation in export participation do not contribute to identification and are omitted, reducing the effective sample from 144 to 129 firms. Clustered standard errors allow for dependence within the 31 sampling strata, but the modest number of clusters can make conventional cluster-robust inference imprecise; for this reason, coefficient signs, conventional significance, CR2-Satterthwaite results, and wild-bootstrap results are reported and interpreted separately.

As additional robustness checks, the article reports: (i) a correlation matrix and variance inflation factors (VIF/GVIF) to assess collinearity; (ii) alternative specifications without fixed effects, with region fixed effects only, and with sector fixed effects only; (iii) penalized Firth logit models as a conservative check given the relatively small sample size; (iv) alternative inference for few clusters using CR2 correction with Satterthwaite degrees of freedom; (v) a wild cluster bootstrap in an auxiliary unweighted specification with the same structure of controls and fixed effects; (vi) models including interactions between managerial practices and quality certification, and between managerial practices and sustainable intensity; and (vii) complementary models using export intensity (export_share) as the dependent variable estimated by means of fractional logit, Tobit, and a two-part approach.

3.6. Interpretation and Identification

Because the coefficients of logit models are expressed in log-odds, interpretation is complemented by average marginal effects and by predicted probabilities of exporting as mp_core and green_index vary between the 10th and 90th percentiles, holding the remaining covariates at their observed values. This strategy facilitates a more substantive reading of the results and avoids restricting interpretation to coefficients that are not especially intuitive.

The study’s identification strategy is associational rather than causal. The cross-sectional design prevents the establishment of temporal sequences among firms regarding management, sustainability, and exporting, and does not allow problems of simultaneity or omitted variables to be ruled out completely. Moreover, although the use of standard errors clustered by stratum is consistent with the sampling design, the relatively small number of strata requires inference to be interpreted with caution. To address this issue, inference is complemented by CR2 correction with Satterthwaite degrees of freedom in a weighted linear-probability analogue of the main model and by a wild cluster bootstrap in an auxiliary unweighted specification. Since these corrections are more conservative, especially with few clusters, they are interpreted as exercises in inferential sensitivity rather than as substitutes for the main model.

4. Results

4.1. Descriptive Overview of the Analytical Sample

Before estimating the econometric models, Table 1 presents the composition of the analytical samples and the main descriptive statistics used in the empirical analysis. Panel A reports the sample sizes available for each block of estimations, showing that the main export sample includes 144 firms distributed across 31 strata, while the effective sample used in the fixed-effects logit models is smaller due to identification requirements. Panel B summarizes the central continuous variables, including managerial practices, sustainability intensity, firm size, age, foreign ownership, and export share. Panel C compares unweighted and weighted indicators for export participation, sustainable adoption, quality certification, process innovation, and managerial practices, highlighting the importance of accounting for the survey design. Finally, Panel D provides an initial descriptive approximation to the relationship between organizational capabilities and exporting, showing higher weighted export probabilities among firms adopting sustainable practices, holding quality certification, or reporting process innovation. These descriptive patterns provide preliminary evidence consistent with the conceptual framework, although they do not imply causal relationships and are formally assessed in the subsequent econometric models.

4.2. Management, Sustainability, and Complementary Capabilities

Before modelling exporting as the main outcome, auxiliary equations were estimated to examine whether managerial practices are associated with sustainability and with specific organizational capabilities. The results are presented in

Table 2. Auxiliary models: sustainability and organizational capabilities.

Variable	M1: green_adopt (Logit FE)	M1b: green_index (OLS FE)	M2a: quality_cert (Logit FE)	M2b: proc_innov (Logit FE)
mp_core	2.552 (2.183)	0.511 (0.313)	3.548 (1.606) *	0.365 (2.008)
ln_emp	−0.131 (0.250)	0.027 (0.037)	−0.100 (0.361)	0.338 (0.302)
age	0.033 (0.011) **	0.002 (0.002)	0.031 (0.024)	−0.011 (0.018)
foreign_share	−0.015 (0.011)	−0.002 (0.001) .	−0.004 (0.007)	0.034 (0.013) **
Observation	173	173	171	146

Notes: standard errors clustered by stratum in parentheses; models with region and sector fixed effects. ** p < 0.01; * p < 0.05; . p < 0.10.

. In the logit model for green_adopt (M1), mp_core displays a positive coefficient, but it is not statistically significant. Similarly, in the linear model for green_index (M1b), the association between management and sustainable intensity is positive, although it also fails to reach conventional levels of significance. By contrast, age shows a positive and significant association with green_adopt, whereas foreign_share exhibits a negative and marginal relationship with green_index. Taken together, these results suggest that management does not automatically translate into observable sustainability within the sample analyzed.

The picture changes somewhat when complementary capabilities are considered. In Table 2, model M2a, mp_core shows a positive and statistically significant association with quality_cert, which is consistent with the view that higher values of the management-practices composite are linked to a higher probability of verifiable standardization. However, in M2b no association is observed between mp_core and process innovation. In this latter case, foreign_share does appear positively and significantly related to proc_innov, suggesting that exposure to foreign capital may facilitate more visible organizational or technological adjustments.

In summary, Table 2 shows that the relationship between management and sustainability is weaker than the relationship between management and exporting, and that managerial practices appear to be linked more clearly to quality certification than to binary sustainability or process innovation.

4.3. Managerial Practices and Export Participation

The study’s main empirical contribution emerges from the export models reported in

Table 3. Main export models.

Variable	M3: Base	M4: +Complementary Capabilities	M5: +Sustainability Intensity	M6: Complementary Capabilities Without green_adopt	M5 LPM
mp_core	4.369 (1.586) **	3.291 (2.709)	4.260 (1.642) **	3.922 (2.518)	0.417 (0.190) *
green_adopt	0.730 (0.815)	0.745 (0.803)	—	—	—
green_index	—	—	1.055 (0.609) .	—	0.151 (0.067) *
quality_cert	—	1.940 (0.909) *	—	1.949 (0.857) *	—
proc_innov	—	1.412 (0.746) .	—	1.387 (0.741) .	—
ln_emp	−1.379 (0.694) *	−1.383 (0.678) *	−1.434 (0.669) *	−1.436 (0.674) *	−0.177 (0.076) *
age	0.052 (0.032)	0.048 (0.032)	0.056 (0.031) .	0.051 (0.032)	0.008 (0.004) .
foreign_share	0.053 (0.022) *	0.049 (0.022) *	0.054 (0.023) *	0.047 (0.022) *	0.008 (0.003) *
Observation	129	129	129	129	144
Adjustment	Pseudo R2 = 0.283	Pseudo R2 = 0.380	Pseudo R2 = 0.287	Pseudo R2 = 0.374	Adj. R2 = 0.244

Notes: cluster-robust standard errors by stratum in parentheses; logit models with region and sector fixed effects, except for the last column, which reports a linear probability model with the same set of controls. ** p < 0.01; * p < 0.05; . p < 0.10.

. In the baseline model (M3), mp_core shows a positive and statistically significant association with the probability of exporting. In this specification, green_adopt is not significant, whereas ln_emp enters with a negative sign and foreign_share with a positive one. This pattern suggests that, once the observed and unobserved heterogeneity captured by the model is taken into account, management provides more consistent explanatory power for export participation than binary sustainable adoption.

When complementary organizational capabilities are incorporated (M4), the coefficient on mp_core declines and loses significance, while quality_cert emerges as a positive and significant predictor and proc_innov as a marginally positive predictor. This attenuation indicates that the management-export association becomes less precise when certification and process innovation are included. It is consistent with the coexistence of these capabilities, but it does not constitute a formal test of causal mediation.

Model M5 replaces green_adopt with green_index, which improves the substantive interpretation of the green dimension. In this specification, green_index shows a marginally positive association, while mp_core remains positive and significant. This evidence suggests that sustainability is linked to exporting more visibly when operationalized as cumulative intensity rather than as a simple binary condition. In other words, it is not sufficient to identify whether a firm reports any sustainable practice; what also matters is the extent to which those practices are expressed cumulatively or with greater intensity.

Model M6, which incorporates complementary capabilities without green_adopt, reinforces the positive and significant association of quality_cert, while proc_innov remains marginally positive. Taken together, Table 3 shows that mp_core has the clearest association among the study’s central variables in the principal specifications, quality_cert is the clearest complementary capability, and green_index provides more informative but weaker and more specification-sensitive evidence than green_adopt.

4.4. Substantive Magnitude and Robustness of the Main Model

To facilitate the economic interpretation of the logit coefficients,

Table 4. Substantive magnitude and robustness of the main model.

Panel (A) Predictions from model M5
Variable	Low Scenario	High Scenario	Export Probability (Low)	Export Probability (High)		Difference
mp_core	p10 = 0.381	p90 = 0.770	0.144	0.357		0.213
green_index	p10 = 0.000	p90 = 1.000	0.193	0.331		0.137
Panel (B) Average marginal effects from Model M5
Variable			AME
mp_core			0.473
green_index			0.117
Panel (C) Robustness of Model M5
Variable	M5 (wmedian)	M5 (wweak)	M5 (wstrict)		M5 GLM Cluster-Robust
mp_core	4.260 (1.642) **	4.534 (1.900) *	3.921 (1.412) **		4.260 (1.635) **
green_index	1.055 (0.609) .	1.015 (0.596) .	1.123 (0.541) *		1.055 (0.606) .
ln_emp	−1.434 (0.669) *	−1.495 (0.702) *	−0.997 (0.540) .		−1.434 (0.666) *
age	0.056 (0.031) .	0.060 (0.030) *	0.044 (0.026) .		0.056 (0.030) .
foreign_share	0.054 (0.023) *	0.053 (0.023) *	0.042 (0.021) *		0.054 (0.023) *
Observation	129	129	129		144

Notes: standard errors clustered by stratum in parentheses. ** p < 0.01; * p < 0.05; . p < 0.10.

, Panel A, presents predicted probabilities associated with relevant changes in the main variables of model M5. When mp_core increases from the 10th percentile (0.381) to the 90th percentile (0.770), the weighted probability of exporting rises from 0.144 to 0.357, equivalent to an increase of 0.213 points. Similarly, when green_index moves from the 10th to the 90th percentile, the weighted probability increases from 0.193 to 0.331, that is, by 0.137 points. These magnitudes suggest that the association between management and exporting is not only statistically relevant, but also substantively important.

Table 4, Panel B, complements this interpretation with average marginal effects. The AME for mp_core is 0.473 and that for green_index is 0.117. Although these values should be interpreted with caution given the non-linear nature of the model and the weighting structure, they indicate a larger estimated change in the probability of exporting for variation in managerial practices than for variation in sustainability. Panel C further shows that the positive signs of the central variables are preserved across the reported specifications. Statistical significance, however, varies across estimators and weighting schemes; accordingly, these exercises support specification stability more clearly than uniform inferential robustness.

4.5. Additional Robustness, Alternative Inference, and Extensions

The additional sensitivity exercises refine the interpretation of the main results rather than establishing uniform robustness. As shown in

Table 5. Collinearity diagnostics and alternative specifications.

Panel (A) Main correlations among regressors
Pair of Variables	Correlation		Pair of Variables	Correlation
mp_core–green_index	0.270		green_index–age	0.237
mp_core–ln_emp	0.227		green_index–foreign_share	0.106
mp_core–age	0.235		ln_emp–age	0.390
mp_core–foreign_share	0.135		ln_emp–foreign_share	0.292
green_index–ln_emp	0.247		age–foreign_share	0.112
Panel (B) VIF/GVIF
Specification	Key Result
VIF without fixed effects	mp_core = 1.01; green_index = 1.11; ln_emp = 1.64; age = 1.29; foreign_share = 1.57
GVIF with region fixed effects	maximum = 1.64
GVIF with sector fixed effects	maximum = 1.88
Panel (C) M5 under alternative fixed-effects structures
Variable	Without FE (144)	FE Region (144)	FE Sector (129)	FE Region + Sector (129)
mp_core	3.146 (1.847) .	3.628 (2.032) .	3.549 (1.989) .	4.260 (1.634) **
green_index	1.710 (0.953) .	1.564 (0.927) .	1.257 (0.9754) .	1.055 (0.699) .
ln_emp	−1.309 (0.899) .	−1.304 (0.988) .	−1.418 (0.888) .	−1.434 (0.688) .
age	0.056 (0.029) . *	0.057 (0.030) . *	0.055 (0.030) .	0.056 (0.031) .
foreign_share	0.052 (0.049).	0.052 (0.049).	0.052 (0.029) .	0.054 (0.029) .

Notes: FE region = a3; FE sector = a4. Errors clustered by stratum in parentheses. ** p < 0.01; * p < 0.05; . p < 0.10.

, Panels A and B, the correlation matrix and the VIF/GVIF values do not suggest severe collinearity problems among the regressors in the main model. The correlations between mp_core, green_index, ln_emp, age, and foreign_share remain in low to moderate ranges, and the variance inflation factors stay below commonly used diagnostic thresholds.

Table 5, Panel C, shows that the specifications without fixed effects and those with partial fixed effects preserve the positive sign of mp_core and green_index. In particular, mp_core moves from a marginal association in the specification without fixed effects to positive associations under the alternative fixed-effects structures, reaching conventional significance when both region and sector fixed effects are included. green_index remains positive and marginal across the four specifications. Complementarily,

Table 6. Inferential sensitivity and coefficient stability.

Panel (A) Sensitivity of the ln_emp coefficient
Model	Coef. ln_emp	p-Value	Model	Coef. ln_emp	p-Value
M3 FE both	−1.379	0.047	M5 without FE	−1.309	0.0297
M4 FE both	−1.383	0.041	M5 FE region	−1.304	0.0311
M5 FE both	−1.434	0.032	M5 FE sector	−1.408	0.0345
M6 FE both	−1.436	0.033	M5 FE region + sector	−1.434	0.0320
Panel (B) Penalized Firth logit models
Variable	Firth Without FE		Firth FE Region	Firth FE Sector
mp_core	1.888; p = 0.162		2.127; p = 0.123	2.598; p = 0.050
green_index	1.433; p = 0.069		1.324; p = 0.104	1.152; p = 0.326
ln_emp	−0.120; p = 0.480		−0.109; p = 0.527	−0.153; p = 0.396
foreign_share	0.0126; p = 0.054		0.0127; p = 0.055	0.0116; p = 0.084
Panel (C) Alternative inference for few clusters
Method	Variable		Estimate/Beta	p-Value
Unweighted auxiliary wild bootstrap	mp_core		0.472	0.045
Unweighted auxiliary wild bootstrap	green_index		0.169	0.091
Weighted CR2 + Satterthwaite	mp_core		0.417	0.788
Weighted CR2 + Satterthwaite	green_index		0.151	0.277

Notes: the wild bootstrap was estimated using an unweighted auxiliary specification with the same fixed-effects structure; CR2 + Satterthwaite was applied to a weighted linear-probability analogue of the main M5 specification.

, Panel A, shows that ln_emp maintains a negative and significant coefficient in all variants considered, suggesting notable stability in this result.

Table 6, Panel B, presents the penalized Firth logit models, estimated as a conservative small-sample check. In these specifications, green_index remains positive, although it reaches marginal significance only in the specification without fixed effects. mp_core also preserves a positive sign, reaching conventional significance only in the sector fixed-effects specification. Thus, penalized estimation preserves the coefficient signs but does not yield uniform statistical significance.

Table 6, Panel C, reports alternative inference for a limited number of clusters. Applied to a weighted linear-probability analogue of the main M5 specification, the CR2 correction with Satterthwaite degrees of freedom produces non-significant estimates for mp_core and green_index (p = 0.788 and p = 0.277, respectively), and therefore does not reproduce the statistical confidence of the principal model. By contrast, the wild cluster bootstrap in the auxiliary unweighted specification yields evidence for mp_core at the 5% level and marginal evidence for green_index at the 10% level. The coefficient signs are preserved across these exercises, but statistical precision depends on the inferential procedure and weighting structure; the evidence should not be characterized as uniformly robust.

Table 7. Interactions and extensions with export_share.

Panel (A) Interactions
Variable	mp_core × quality_cert		mp_core × green_index
mp_core_c	8.024; p = 0.0058		4.215; p = 0.0095
quality_cert/green_index_c	2.754; p = 0.0045		0.848; p = 0.200
Interaction	−15.311; p = 0.021		5.539; p = 0.213
Panel (B) export_share as the dependent variable
Model	mp_core	green_index	ln_emp	foreign_share
Fractional logit	2.919; p = 0.132	1.011; p = 0.009	−0.402; p = 0.363	0.025; p = 0.025
Tobit	0.609; p = 0.048	0.235; p = 0.055	−0.015; p = 0.703	0.0038; p = 0.006
Two-part 1: exports	4.260; p = 0.009	1.055; p = 0.082	−1.434; p = 0.031	0.054; p = 0.019
Two-part 2: intensity	0.204; p = 0.917	0.398; p = 0.499	0.028; p = 0.895	0.0057; p = 0.501

Notes: p-values reported from the complementary specifications with export_share as the dependent variable.

, Panel A, summarizes the interaction models. The interaction between mp_core and quality_cert is negative and statistically significant, whereas the interaction between mp_core and green_index is not statistically significant. The negative interaction coefficient indicates a non-additive pattern within the estimated specification, but it does not identify the underlying mechanism. One possible explanation is that certified firms may already embody some of the organizational routines captured by the management-practices composite. However, selection into certification, reverse relationships between exporting and certification, differences in firm composition, and unobserved organizational characteristics are also compatible with this pattern. The interaction should therefore be interpreted as suggestive rather than as evidence of a specific mechanism.

Finally, Table 7, Panel B, presents the complementary exercises in which export_share is used as the dependent variable. Sustainable intensity has a positive coefficient in the fractional logit model. In the two-part approach, the central variables are more informative for the decision to export than for export intensity among firms that already export. This pattern indicates that the observed organizational capabilities are more clearly associated with the extensive margin of export participation than with the intensive margin of foreign sales.

5. Discussion

Across the principal specifications, mp_core displays the clearest association with export participation, although its precision weakens under some conservative procedures. Substantively, this pattern is compatible with management as an organizational capability for coordinating and adapting firm operations, rather than solely as an instrument of internal efficiency. This reading accords with the literature linking managerial practices to persistent differences in firm performance and organizational adaptability in competitive environments (Bloom & Van Reenen, 2007; Quinn & Woodruff, 2021; Scur et al., 2021).

Quality certification provides the clearest evidence among the complementary capabilities, while process innovation shows a more moderate and specification-sensitive association. When these variables are incorporated, the coefficient on mp_core becomes smaller. This attenuation indicates that management, verifiable standards, and operational improvements coexist in the empirical model, but it does not establish that certification or innovation mediates the management-export relationship. The pattern accords with evidence linking certification to export outcomes and with research on process innovation in developing economies (Yang et al., 2023; Edeh et al., 2020; Najafi-Tavani et al., 2025).

Sustainability also warrants a nuanced interpretation. green_adopt is not statistically significant, whereas green_index is positively associated with export participation in the main model. The relationship is therefore more visible when sustainability is measured as the cumulative intensity of practices rather than as a binary condition. This contrast highlights the importance of measurement depth but does not establish that increasing sustainable practices would improve export performance. The interpretation is consistent with research showing that the relationship between green strategies and export outcomes depends on readiness, innovation, and complementary capabilities (Singh et al., 2024; Cai & Hao, 2025).

The sensitivity analyses qualify these empirical patterns. The signs of mp_core and green_index are generally preserved, but the conservative few-cluster procedures yield different levels of statistical precision. The discussion therefore focuses on the direction and relative salience of the associations rather than on uniform statistical significance.

The negative interaction coefficient between mp_core and quality_cert indicates a non-additive pattern in the estimated specification, but it does not identify why that pattern arises. One possible explanation is that certified firms may already embody some of the organizational routines captured by the management-practices composite. Nevertheless, selection into certification, reverse relationships, differences in firm composition, or other unobserved organizational characteristics could produce a similar result. The interaction therefore motivates further investigation rather than establishing a preferred substantive mechanism. By contrast, the interaction between mp_core and green_index is not statistically significant.

Taken together, the findings support export readiness as an integrative lens for interpreting how differentiated organizational capabilities coexist around export participation. They do not support treating all capabilities as equivalent: managerial coordination and verifiable standardization display the clearest evidence, whereas sustainability intensity and process innovation are less precise and more specification-sensitive. For emerging economies, this differentiated pattern suggests that organizational readiness may complement structural explanations based on firm size, age, and ownership. The positive association of foreign_share further indicates that external linkages may coexist with internal capabilities (Vendrell-Herrero et al., 2025). These interpretations remain subject to the cross-sectional design, the limited measurement of unobserved organizational characteristics, and the relatively small number of exporting firms. Accordingly, the contribution lies in organizing differentiated evidence on internal capabilities around export readiness, not in identifying a causal bundle.

6. Conclusions

In the principal specifications, managerial practices display the clearest association with export participation, while quality certification is the clearest complementary capability. Sustainability intensity shows a weaker, measurement-sensitive association, and process innovation provides less precise evidence. This differentiated pattern is consistent with a multidimensional view of export readiness in which internal coordination, verifiable standardization, operational adaptation, and cumulative upgrading represent distinct, but not equally supported, organizational dimensions. This perspective complements structural explanations of exporting without implying that any single capability causally determines export participation.

The additional sensitivity analyses generally preserve the signs of the central coefficients, although statistical precision varies across procedures. Consequently, the conclusions concern patterns of association rather than uniformly robust or causally identified effects.

From a public policy perspective, the findings identify areas for future experimentation rather than interventions whose effectiveness can already be assumed. The associations involving managerial practices and quality certification motivate the evaluation of support for managerial routines and verifiable standards, while the contrast between green_adopt and green_index suggests that sustainability initiatives should assess the intensity and continuity of upgrading. Process improvement may also be included in such pilots, but its less precise association warrants particular caution. Pilot programs, phased implementation, and prospective impact evaluations are needed to assess effects on export entry and export intensity.

The study opens avenues for future research. Longitudinal or quasi-experimental strategies would help distinguish the temporal sequence between management, sustainability, certification, and exporting. Future work should also examine the sectoral and territorial heterogeneity of these associations. Within this agenda, the evidence presented here provides an empirical starting point for understanding how internal organizational capabilities are linked to the international integration of firms in emerging economies.