Due to the resource constraints that SMEs often face, international experience is considered essential for the international expansion of a company (Andersson, 2014), while the lack of such experience is considered an obstacle and a major barrier (Velez-Ocampo, Gonzalez-Perez, & Sin, 2021). International business experience in different geographical locations and contexts fosters the creation of international contact networks (De Cock et al., 2021). It has been asserted that entrepreneurs dealing with international markets seem to learn from such experiences and accumulate valuable knowledge about doing business in foreign markets, which facilitates the recognition of export opportunities (Sandoval-Álvarez, 2023). Previous international experience and skills developed in serving foreign markets can explain the propensity of managers to develop their businesses internationally (Loué, 2018).

According to IE scholars, international business experience can increase an individual’s willingness to seek international ventures (Li & Liu, 2018). The more time spend interacting in foreign markets, the more knowledge is gained about successful international procedures and businesses, making them more efficient and effective (Evangelista & Mac, 2016). Previous research has suggested a positive impact of international experience on the international performance of firms, such as export intensity and propensity (i.e., Rubino, Vitolla, & Garzoni, 2017; van Staden, 2022). Consequently:

H1: The international experience of a SME key manager is positively related to the degree of internationalization of the SME.

Other types of experience, not exclusively related to international business experience, may be relevant for successful internationalization. In the IE literature, this is referred to as international orientation, global mindset, or international mindset (He, 2020). It can be enhanced by a range of personal experiences, not necessarily related to business experience. Global mindset is rooted in the knowledge gained through vast international travel experience (He, Baranchenko, Lin, Szarucki, & Yukhanaev, 2020), living experience abroad (Andresen, & Bergdolt, 2021), foreign language proficiency (He et al., 2020). It is considered an intangible, idiosyncratic, and difficult-to-trade capability (Kyvik, 2018). Exposure to other cultures and international contexts through living abroad, and regular international travel leads to a strong international orientation. Managers who have had the experience of living abroad are more likely to respond positively to export opportunities than their counterparts without this experience and are more receptive to the stimuli of doing business abroad (Andresen, & Bergdolt, 2021).

Having a global mindset fosters a generally positive attitude towards international affairs and facilitates the ability to adapt to international environments and cultures (Lazaris, & Freeman, 2018). It has been shown that internationally oriented managers are more likely to succeed in the internationalization process of SMEs (e.g., Torkkeli, Nummela, & Saarenketo, 2018). Previous studies confirm that SME managers with higher levels of international orientation are more prompt to adopt an export strategy faster than firms managed by individuals with low international orientation (Galati et al., 2014). They are better identifying business opportunities in foreign markets (Anschütz,& Mazzucato, 2022). For internationally oriented managers, choosing a strategy for entering new international markets is less difficult than for their counterparts without such orientation (Nielsen & Nielsen, 2011). Having an international orientation also encourages both international networking and knowledge acquisition activities, which in turn affects the development of international opportunities positively (He, et al., 2020). Finally, according to research conducted by Kyvik (2018) on SMEs, the most significant finding of his research is the confirmation of a strong causal effect of the international mindset on SMEs’ internationalization (Kyvik, 2018).

Based on the above, it seems plausible that internationally oriented managers are more alert to international opportunities and more likely to take advantage of them once they are identified. Accordingly:

H2: The international orientation of a SME key manager is positively related to the degree of internationalization of the SME.

Business internationalization processes expose managers to new and unfamiliar contexts or business challenges in various facets (Freixanet & Renart, 2020); for example, limited access to market information, new regulations, trade barriers, etc. Low familiarity with the foreign market increases the perception of risk, which in turn reduces the likelihood of the firm’s international expansion (Fornes & Cardoza, 2018). The lack of information and knowledge about the target markets abroad, as well as political instability, increases risk and can negatively affect the manager’s decision to expand internationally, which in turn would affect the export performance of the company (Mathews, Maruyama, Sakurai, Perks, & Sok, 2018).Thoughts and perceptions play a fundamental role in the internationalization of firms, as managers thoughts determine their behavior (Sandoval & Rank, 2022). Risk perception can motivate or prevent an individual from taking certain actions (Handrito et al., 2020) because it is a personal assessment of the likelihood of negative outcomes (Wolff, Larsen, & Øgaard, 2019) It has been argued that managers with more business experience at the international level perceive less risk with respect to exports. The opposite also seems to be the case; that is, managers with little international experience do not often consider exports as a strategy for firm growth (Fornes & Cardoza, 2018).

Previous research on SME firms shows that in-ternational experience acquired by a manager increases their sense of controllability of international ventures, which in turn reduces the sense of uncertainty inherent in foreign markets (Eduardsen & Marinova, 2020). Therefore, it seems plausible to assume that:

H3: The level of international experience of a SME key manager is negatively related to their risk perception regarding export.

According to He et al. (2020), having an internationally oriented mindset has a profound effect on pursuing international business opportunities. People who have been exposed to foreign markets and diverse cultures for a significant period, whether through extensive travel or living abroad, create a sense of familiarity with countries and their specific markets (Kyvik, 2018).

Some authors also argue that internationally oriented managers perceive less ambiguity and complexity when seeking opportunities in foreign markets (Sandoval-Álvarez, 2019). It was also shown that globally oriented managers positively impact the international opportunity identification in foreign markets (He et al., 2020). A higher level of international orientation among SME managers can reduce the effect of culture shocks and increase knowledge of business practices in other countries (Zhou et al, 2022). Previous experiences from living or working abroad develop an internationally oriented mindset in these individuals, reducing the liability of novelty, that is, the perception of unfamiliarity with foreign countries and business practices (Fornes & Cardoza, 2018), and reducing the perception of "psychological distance" from such markets (Jones & Casulli, 2014). The above leads us to postulate that:

H4: The level of international orientation of a SME key manager is negatively related to their risk perception regarding export.

According to some IE scholars, assessing risks is an integral part of evaluating the inherent threats associated with international business (Milanov & Maissenhälter, 2014). The perceived uncertainty of some foreign markets affects the level of effort required to enter them (Guercini, Milanesi, & Runfola, 2020). Studying individual entrepreneurs’ perceived risks and barriers in foreign markets may help explain why some firms opt for exporting while others remain focused on domestic markets (van den Berg, 2022). In theory, expanding export activities requires processing numerous opportunities and threats, leading to information overload for managers. They are unable to examine all available information under these circumstances. Instead, they tend to rely on their perceptions to deal with the novelty and complexity that characterize the export business (Jones & Casulli, 2014). The decision to expand into foreign markets is based on the decision-maker's perception of risk and their ability to manage uncertainty (Guercini et al., 2020). Milanov and Maissenhälter (2014) emphasize the importance of understanding how managers deal with risk and uncertainty that arise when exporting in order to comprehend the international entrepreneurial behavior of a firm.

Empirically, some research has demonstrated that risk perception highly correlates with a company’s inter-nationalization outcomes as it is considered a significant precursor that shapes the internationalization processes and influences the scope of the export operations (da Silva & Takahashi, 2021). Previous studies indicate that SMEs with managers who perceive lower levels of risk tend to be more involved in export operations (da Silva & Takahashi, 2021). Consequently:

H5: A SME key manager’s perception of risk regarding exports is negatively associated with the degree of internationalization of their SME.

Our model describes the links between the predictor variables and proposes that risk perception may act as a mediating variable. The following theoretical arguments from the IE literature are offered to support this hypothesis.

In their review of EI studies, Keupp and Gassmann (2009) found a lack of research exploring the links between variables that predict the international performance of a firm, referred to as the "black box" of international entrepreneurship research. Faroque et al. (2020) have argued that prior experience may not directly impact internationalization performance due to inconsistent results in previous studies. As mentioned earlier, some IE academics hold that understanding the search for international companies requires considering risk assessment capacity (Acedo & Jones, 2007).

Sandoval and Rank (2022) contend that managers who perceive themselves as experienced or skilled see the environment as safe, while those who consider them-selves less capable perceive more threats. Carpenter et al. (2004) point out that international expansion of SME export activities is a risky endeavor that involves obtaining and processing information on market conditions, which can lead to information overload for managers and further uncertainty and ambiguity. Under such circumstances, managers can only filter a limited portion of available information and interpret it based on their own perceptions.

Managers engage in a cognitive activity where they filter and interpret situations based on their perceptions of familiarity and novelty (Sandoval-Rank, 2022). The perception of a manager is considered to be a cognition element that may play an intervening role. In this vein, Milanov & Maissenhälter (2014) have argued that cognitive factors could “add refinement and layers of complexity to the mechanisms that mediate how the founders’ personal internationalization experience informs the firm’s subsequent internationalization behavior” (p. 50). According to Jones and Casulli (2014), cognition elements precede international behavior and performance. Similarly, Faroque et al. (2020) argue that knowledge gained in global settings should first be applied to the cognitive thinking of the entrepreneur. From these statements, we could conclude that managerial perceptions play a mediating role in the relationship between managers' international orientation and experience and companies outcomes. Therefore, it is possible to assume that:

H6 a): The relationship between the international experience of a key manager and the degree of internationalization of their SME is mediated by the key manager’s perception of risk regarding exports.

H6 b): The relationship between the international orientation of a key manager and the degree of internationalization of their SME is mediated by the key manager’s perception of risk regarding exports.

Covariance-based Structural Equation Modeling (CB-SEM) was used to test the hypotheses. This technique estimates the strength of the hypothesized relationships represented in the model at the same time, making clear the relative significance of each relationship in the model (Hair, Black, Babin, & Anderson, 2010).

To ensure the accuracy of the hypothesized research model using CB-SEM, it is imperative to thoroughly evaluate the reliability and validity of the model's constructs, as recommended by Hair et al. (2010). The validity and reliability of the constructs were verified prior to analyzing the model and confirming the hypotheses. To verify the multi-item internal consistency of the reflective variable (Hair et al., 2010), the composite reliability was evaluated by computing the C.R. coefficient. Values of C.R. greater than 0.7 evidence composite reliability. The risk perception (reflective variable in the model) had a C.R. value of 0.81, which indicates consistent items within this construct.

Convergent validity was checked by analyzing the extent to which items of a variable correlate with each other (Hair et al., 2010; Byrne, 2016). Confirmatory factor analysis (CFA) was conducted using the Amos28 software to test the significance of item loadings that belong to the reflective variable in the model (Hair et al., 2010). All item loadings were statistically significant at p<.001. In addition, we calculated the Average Variance Extracted (AVE) to determine if the standardized item estimates were above their acceptable threshold (i.e., AVE>0.50). The AVE value of the Risk Perception variable was calculated to be 0.59. These results indicate that the items included in this construct share a significant amount of variance, demonstrating convergent validity.

Discriminant validity, the degree to which the independent variables in the model are truly different from each other, was also tested (Hair et al., 2010). For this purpose, the Heterotrait-Monotrait Ratio of Correlation (HTMT) coefficients were estimated. According to Henseler et al. (2015), there is evidence of discriminant validity between constructs when the HTMT coefficients are below the threshold of 0.90 for each pair of constructs examined. When assessing the discriminant validity between the three predictor variables, the HTMT coefficient resulted in values of -0.436, -0.352, and 0.290, indicating that the predictor variables of the model are different from each other, thus demonstrating discriminant validity. In addition, the assumption of non-multicollinearity between the independent variables was tested. High intercorrelations above 0.80 between predictor variables would indicate multicollinearity problems and a potential threat to the study results (Hair et al., 2010). However, all correlations between independent variables were well below the threshold, ranging from -0.33, -0.26, and 0.29. To strengthen this test, variance inflation factors (VIFs) were calculated. The VIF values ranged from 1.076 to 1.27. According to Hair et al. (2010), VIF values below the threshold of 5 indicate non-redundancy among the predictor variables.

Finally, the formative constructs in the model were evaluated, specifically the degree of internationalization and international orientation. The formative variables are composite variables formed by the sum of their dimensions (Hair, Ringle, & Sarstedt, 2011). Therefore, evaluating internal consistency, reliability, and convergent validity is not appropriate. For formative variables, the primary concern is the high level of multicollinearity among their dimensions. Multicollinearity was examined to confirm that the two dimensions of Doi, namely international export sales ratio and number of countries, are not highly correlated. It also applies to the aspects of international orientation, such as the number of languages spoken or the likelihood of traveling, living, or working abroad.

To examine multicollinearity, we calculate the VIF factor (Variance Inflation Factor). VIF values greater than 5 generally indicate multicollinearity problems (Hair et al., 2010). The two dimensions of Doi showed VIF values of 1.012, while the two dimensions of a manager's international orientation showed VIF values of 1.076, 1.092, and 1.112. These results suggest that multicollinearity is not a concern. The results of the consistency, reliability, convergent and discriminant validity tests are summarized in Appendix 2.

The goodness-of-fit was evaluated by calculating the coefficient of minimum discrepancy (χ2/d.f.) whose value was 1.076. It is less than the maximum threshold of 3.0 (e.g., Bryne, 2016); this suggests an acceptable model fit. Because chi-square (χ2) values are susceptible to changes in sample size and model complexity, a series of additional indices were checked to make better judgments about the model goodness-of-fit. Following Weiber and Mühlhaus (2014), five indices were analyzed. Table 1 summarizes these index scores and respective thresholds. Overall, the indices show that the model fits the data quite well.

Table 1

Fit Index	Thresholds	Scores
Tucker- Lewis Index (TLI)	≥ 0.95	0.989
Incremental Index of Fit (IFI)	≥ 0.95	0.996
Comparative Fit Index (CFI)	≥ 0.95	0.996
Relative Non-Centrality Index (RNI)	≥ 0.95	0.995
Root Mean Square Error of Approximation (RMSEA)	≤ 0.06	0.024

Notes: Fit indices y thresholds according to Weiber & Mühlhaus (2014).

Lastly, the assumption of multivariate normality in the data was verified. The multivariate non-normality in the data could be detrimental to Covariance-Based Structural Equation Modeling analysis (CB-SEM) because it may lead to biased statistics and attenuated standard errors (Awang, Afthanorhan & Asri, 2015). Multivariate normality was checked by computing Mardia’s normalized coefficient (d2) as suggested by Byrne (2016). This coefficient yielded a value of d2= 9.18. Mardia values less than 5 indicate the presence of multivariate normality (Byrne 2016). Consequently, the data failed to meet the assumption of multivariate normality in the data.

In order to handle the multivariate non-normal data, we used bootstrap procedures, often used in cases where the assumption of multivariate normality is not met. Specifically, a Bollen-Stine bootstrapping was performed to test the model fit. This is considered as an alternative effective solution to approximate the sampling distribution of a statistic when the population distribution is not normal or unknown (Ferraz, 2020). Following Byrne (2016), first, a Bollen-Stine bootstrap test was run to check whether the p-value of the chi-square of fit statistic is greater than 0.50. The result, p= 0.16, provided evidence of good model fit. Second, once the suitability of our model was confirmed and a second bootstrap procedure with 1,000 interactions was applied to obtain the parameters estimates and the stability of their p-values (Nevitt and Hancock, 2001). It was demonstrated that 1,000 interactions are enough to obtain high accuracy of the parameters estimates.

The results of the bootstrap procedure indicate that the relationship between international orientation and the degree of internationalization was not significant (β= 0.01, p>0.05), therefore rejecting hypothesis H1. The relationship between international experience and the level of internationalization was found to be insignificant (β= 0.11, p>0.05), leading to reject hypothesis H2. Regarding the relationship between international orientation and risk perception towards export expansion, it was found to be significantly negative (β= -0.21, p<0.01); hence, Hypothesis H3 is supported. As hypothesized, there is a negative and statistically significant relationship between international experience and risk perception towards exports (β= -0.31, p<0.01); thus, hypothesis H4 is supported. Additionally, the results indicate that risk perception towards expanding exports is negatively correlated with the level of internationalization (β= -0.43, p<0.00), supporting hypothesis H5. Figure 2 summarizes the results of the hypotheses tested.

Figure 2 Notes: Only relationships highlighted in blue color are statistically significant. The thickness of the arrow represents the relative weight of the β value. **= statistically significant p< .001; *= statistically significant p< .05; η= not statistically significant Source: own elaboration.

To check the stability of the p-values and β parameters, a bootstrap procedure was carried out by applying 3,000 interactions. This test allows us to make a more reliable assessment of the statistical significance of the standard errors and the β estimates (Byrne 2016). The final results are shown in Table 2. The β estimates and p-values remain unchanged and significant, supporting hypotheses H3, H4, H5.

Table 2

	Hypothesized Relationships				Lower β bound	Upper bound	p b
International orientation	→	Degree of internationalization	H1	0.016	-0.176	0.177	0.930
International experience	→	Degree of internationalization	H2	0.107	-0.107	0.311	0.395
International orientation	→	Risk perception	H3a	-0.209	-0.397	-0.015	0.039
International experience	→	Risk perception	H4a	-0.307	-0.445	-0.170	0.002
Risk perception	→	Degree of internationalization	H5a	-0.429	-0.602	-0.276	0.001
Industry branch	→	Degree of internationalization	C1	0.065	-0.142	0.240	0.582
Company’s size	→	Degree of internationalization	C2	0.230	-0.094	0.489	0.141
Company’s exporting age	→	Degree of internationalization	C3	0.092	-0.146	0.324	0.440
Governmental support	→	Degree of internationalization	C4	0.052	-0.130	0.254	0.557

Notes: ^a = supported relationships ^b = p-value.

A bootstrap procedure was conducted in accordance with the recommendation of Hair et al., (2010) to test for mediating effects. First, we set the indirect paths of the model to zero to compute the β-estimates and corresponding p-values of the direct paths. Subsequently, we applied the same bootstrap procedure to the model, but this time without constraining the effect of the indirect paths to zero. A mediating effect is demonstrated when a change in the strength of the statistically significant relationship between the direct path of the model is significantly reduced or even reduced to zero when an indirect path is included, that is, by introducing an intervening variable into the model (Field, 2013). The results of these tests are summarized in Tables 3 and 4.

Table 3

Paths	β-estimates	Confidence intervals		p-values	Significance
		Lower bound	Upper bound
Direct path: 1)	0.339	0.193	0.460	0.003	supported
Direct Effect:
Indirect path: 2)
Direct Effect:	0.155	-0.015	0.294	0.069	not supported
Indirect Effect:	0.149	0.078	0.280	0.001	supported
Total Effect:	0.304	0.152	0.439	0.003	supported

Notes: 1) Results when the indirect path in the model is constrained to zero. 2) Results when indirect path in the model is not constrained to zero

Table 4

Paths	β-estimates	Confidence intervals		p-values	Significance
		Lower bound	Upper bound
Direct path: 1)	0.210	-0.025	0.364	0.070	not supported
Direct Effect:
Indirect path: 2)
Direct Effect:	0.020	-0.201	0.187	0.888	not supported
Indirect Effect:	0.101	0.014	0.211	0.025	supported
Total Effect:	0.121	0.101	0.283	0.284	supported

Notes: 1) Results when the indirect path in the model is constrained to zero. 2) Results when the indirect path in the model is not constrained to zero

In summary, the results confirm that a key manager’s risk perception mediates the relationship between a key manager’s international experience and degree of internationalization. The strength of the direct path is reduced by non-significant (β= 0.155, p >0.05) when the indirect effect is not constrained; therefore, hypothesis H6 a) is supported.

Finally, contrary to our expectations, hypothesis H6 b) is not supported. There was no evidence of a direct effect of international orientation and the degree of internationalization of an SME when the indirect effect is not restricted.