Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. Summary and General Assessment

The manuscript, which is trying to link digital skills with different forms of entrepreneurship in Mexico, it relies on a very large micro dataset from the 2021 Digital Skills Profiler, including more than fifty thousand observations, and it estimates three probit type models that separately consider microenterprise owners, their motivation (opportunity versus necessity) and also individuals who have not yet but may become entrepreneurs, and the authors conclude that women and older people lag in digital capabilities, while education is strongly positive, however, paradoxically people with high digital skills are less likely to create firms because they prefer to enter stable salaried jobs instead, and if they do start ventures they are concentrated in service or unconventional sectors rather than goods trade, which is a result that appears interesting but also raises methodological and conceptual doubts.

Overall I consider the topic important and policy relevant, but due to the inconsistencies in variable definitions, contradictions in descriptive statistics and the way regression results are interpreted, in its present form the manuscript does not provide reliable evidence.

2. Section by Section Comments

(A) Title, Abstract, Keywords

• The abstract promise to test the effect of digital skills on entrepreneurship (L14–21), but in reality the dependent variable is different, since Model 1 and 3 use dig-skills and Model 2 use motivation, so the description is misleeding, and unless the authors include a model with “entrepreneurship=1” in the whole sample the claim is overstated.

• In the same section the conclusion that “advanced skills reduce entrepreneurship” (L16–23) is too strong, because without that core model this cannot be infered directly, so please either add the missing regression or soften the statement.

• The keyword list has duplication, technological gap appear twice (L30–31), which is not appropriate, also terms like occupational choise or entrepreneurial selection would be more relavant.

(B) Introduction

• Hypothesis and tests do not really match, as in L72–76 you say “advanced skills do not foster entrepreneurship” but in section 4.3 (L405–409) you write that the dependent variable is being entrepreneur, which is not the case in your models, and this misalignment confuse the reader.

• There is also a typo “contributes to a broder understanding” (L78–82), should be “broader”.

• Important methodological details are hidden in footnotes (after L82), making reading flow broken, please bring them up to text.

(C) Background

• The classification into basic/advanced/frontier skills (L118–125) is not explained enough, no mapping of survey items is provided, without that replication is impossible.

• DMI country comparisons (L190–196) may not be comparable, it would be bettter to show if the indicators are harmonized.

• The timeline of policy measures (Table 1, L248 onwards) and the survey years 2020–2021 (L298–307) are not clearly linked, which is problematic because policy impact cannot be traced.

(D) Descriptive

• The paper first says only 3.1% of people without skills (L52–55), but later claims 49.1% of entrepreneurs have no skills (L354–356), which is impossible to reconcile, so numbers must be checked again.

• Also there is confusion on geography: you say location not available for all (L568–571), but earlier you compare states like Chiapas or Oaxaca (L260–276), this contradiction must be resolved, otherwise readers will suspect.

• Variable of “social media use” is not clear, and in Model 3 it has negative effect (L498) which is counterintuitive, so you must define more precise.

(E) Data and Methodology

• The dataset is self–selected, not probabilistic (L298–307), and no weights are applied, which means external validity is weak, please at least show comparision with INEGI benchmarks.

• Variables dig-sk, comp-use, net-use (L460–476) are very colinear, you should check VIF or do PCA, otherwise interpretation overlap.

• The note in L436–439 says “nonsignificant variables deleted”, this practice is not acceptable, because it create bias, so you need to present full models.

• Section 4.3 (L405–409) says dependent variable is entrepreneur, but actually it is not, again inconsistent.

• For age, the marginal effect is reported “per 15 years” (L441–447; L493–499), but this scaling not explained, which confuse readers.

• Coding of opportunity vs necessity entrepreneurship (L458–476) is not fully explained, the survey items should be provided.

• In general, the most serious gap is the absence of a core entrepreneurship regression, without it, main claims lack solid evidence.

(F) Results and Discussion

• Interpretation is too strong. Model 2 show negative relation between digital skill and opportunity entrepreneurship (L465–476), but you explain it as high skilled people preferring labor market, which may be true but it could also be endogeneity. No robustness is done.

• Sector heterogeneity is not tested (L452–455), grouping is too rough.

• The negative coefficient for social media (L500–504) is left unexplained. It may be due to non-strategic usage, but analysis missing.

(G) Conclusions and Policy

• The causal language in conclusions (L540–552) is excessive. This study is cross sectional correlation, so you cannot claim causality.

• Policy recommendations are very general, they should be differentiated between necessity and opportunity entrepreneurs, and by gender or age groups.

(H) Data Availability

• Inconsistent statement: conclusion says data from 2018–2023 reports (L568–571) while main text says 2020–2021 DSP microdata (L298–307). Must be harmonized.

• Only “Gretl 2025a” is mentioned (L425–426), but no scripts shared, so replication impossible.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

This paper is strong and important. It studies digital skills and entrepreneurship in Mexico with a very big dataset and clear method. The results are new and interesting, and they give value for both researchers and policymakers. The writing is clear, the tables are good and the topic is very relevant for sustainability goals.

The background part is detailed and shows good knowledge. But it is also very long and sometimes too descriptive. It would help the reader if this part is shorter and more direct, so that the focus stays on the results and the contribution of the paper.

The results are well presented and very clear. The paradox that people with advanced digital skills are less likely to be entrepreneurs is very original. Still, this idea needs more discussion. The reader wants to know more about why this paradox happens and what it means for society. For example, is it because people with digital skills prefer safe jobs with higher pay, or because they face barriers to start their own firms? Please explain this more.

The policy part in the conclusion is good, but it can be more concrete. It would be better if the paper includes more practical recommendations, like programs to support women and older people with digital tools, or incentives for skilled people to stay in entrepreneurship. These examples will make the paper more useful for decision makers.

The tables are very clear, but the paper could also use one or two figures or simple graphs. This would help readers to see the main findings more quickly.

It is also good to say more about the limits of the study. The data are based on self-reports, and this can sometimes change the accuracy. A short note on this will make the paper stronger and more transparent.

This is a very good and well-prepared paper. It is ready for publication, but it needs some small changes in the discussion, in the explanation of the paradox and in the policy recommendations.

Author Response

Dear reviewer, 

Thank you for your comments; we really appreciate your review. We have addressed your observations on a shorter and more direct background. Also, the reason of the paradox is now mentioned and explained. 

The public policy recommendations are now presented in a more descriptive way, replacing the previous table, in order to be more clear for the reader as you suggested.

Reviewer 3 Report

Comments and Suggestions for Authors

This is a valuable study that addresses a critical intersection of digitalization, entrepreneurship, and policy in an important emerging economy. The use of a large-scale dataset from the Digital Skills Profiler provides a strong empirical foundation. My comments below are intended to help strengthen the manuscript for publication.

1. The most important finding of this study is the "paradox" that individuals with advanced digital skills are less likely to pursue entrepreneurship. The authors convincingly argue this is because these individuals can secure better-paid and more stable employment in the formal labor market. While this is a plausible and fascinating conclusion, the paper could explore its implications more deeply. Rather than simply a paradox, this finding could be framed as a rational economic choice reflecting the specific structure of the Mexican labor market and the nature of its MSME sector. The current analysis suggests a potential market failure: the entrepreneurial ecosystem may not be robust enough to adequately reward and attract high-skilled digital talent. The authors could strengthen their argument by discussing this more explicitly in the conclusion. Are the returns to digital skills in entrepreneurship (especially in its early, necessity-driven stages) so low compared to formal employment that they create a "skills drain" away from new venture creation? A more direct framing of this issue would elevate the paper's contribution from an interesting empirical observation to a powerful commentary on the structural challenges facing Mexican entrepreneurship.
2. The econometric analysis is robust, but a few points regarding the models and variables could be clarified to enhance transparency and reader confidence. First, the dependent variable in Models 1 and 3, "possession of digital skills," is dichotomized into a binary variable (0 for none/foundational, 1 for basic/advanced). However, the original variable appears to be ordinal with four categories. This dichotomization could result in a loss of valuable information. The authors should consider justifying this choice more thoroughly or, alternatively, re-estimating these models using an ordered probit specification. This could yield a more nuanced understanding of the determinants of skill acquisition at different levels. Second, a few key variables could be defined more precisely. The entrepreneurial motivation variable (biz-mot) is central to the analysis in Model 2, yet its construction is not detailed. It would be helpful for the authors to explain exactly how this was measured or what survey questions were used to create this binary classification. Similarly, the main independent variable in Model 2, dig-sk, is also presented without detail on its construction in that specific model, which is critical given its counterintuitive negative coefficient.
3. The manuscript provides a solid review of digital transformation concepts and the relevant policy background. However, the paper's theoretical contribution could be significantly enhanced by a deeper engagement with the academic literature on digital innovation and entrepreneurship. This field often explores how new technologies create opportunities and are expected to drive the formation of competitive new ventures. The paper's central, paradoxical finding offers a valuable opportunity to contribute to, and perhaps even challenge, this existing conversation. A more developed literature review would allow the authors to better position their results against current theories and assumptions in the field. This would help to more clearly articulate the novelty of the study's findings and broaden its relevance to an academic audience focused on the intersection of technology, innovation, and entrepreneurship.
4. The discussion of policy implications is a major strength of the paper. The call to distinguish between policies that support subsistence ventures and those that foster innovative, competitive entrepreneurship is spot-on and well-argued. However, the connection between the specific empirical results and these recommendations could be made even more explicit. The conclusion that digital skills training "may, under certain conditions, inadvertently reduce entrepreneurship rates" is a powerful and provocative claim that follows directly from the paper's main paradoxical finding. To make this point land with maximum impact, the authors could more directly link it back to the specific results. For example, they could highlight how the negative coefficients on biz-mot in Model 1 and dig-sk in Model 2 are the statistical evidence for this potential policy outcome. This would demonstrate clearly that their policy warning is not just speculation but a direct interpretation of their econometric models, thereby strengthening one of the most novel arguments in the manuscript.

 

Author Response

Please see attachment

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Everything's ok; the authors have addressed all my previous concerns about the manuscript. Therefore, I recommend accepting and publishing the manuscript immediately.