The Future of Africa’s Digitalisation: Evidence from Phillips–Sul Convergence Clubbing and Predictive ML Models

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This is a very good attempt to use ML for predictive analytics specific to Africa. Although it is a good study, I have a few concerns –

1. The abstract needs to be rewritten, focusing more on the key outputs from the study. The abstract is missing all numeric values, which is actually helpful to draw inferences. Avoid generic sentences.
2. “The results highlight that digitalisation is a systematic and continuous process driven by the advancement of digital technologies and countries’ capacity to adopt these technologies.” This is obvious, isn’t it? Anybody can get this without any ML.
3. The introduction looks good, but the knowledge gap is not clear. Please ensure the knowledge gap is clearly presented.
4. In materials and methods, describe the Phillips and Sul's convergence and club clustering algorithm. Also, provide a pictorial overview of the methodology.
5. Table 3 can be removed. The info can be placed somewhere else as text.
6. The study is focused on Africa; the dataset used is global. How is this diversity in the dataset attributed in the study? There is no strong argument on that.
7. Some of the results may be put in the appendix, for example, PCA results.
8. The discussion section needs to be revamped, outlining key achievements for Africa. Also, policy interventions and/or policy suggestions should be provided.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

This study employs the Phillips-Sul convergence club algorithm and machine learning models to explore Africa’s digitalisation trends, with a focus on convergence patterns and key driving technologies. I have the following comments, which may help strengthen the manuscript.

1. The research gap section could be more targeted. The current description only mentions the lack of attention to technology-specific convergence and temporal dynamics, but fails to clearly distinguish how this study advances beyond recent key literature. It is recommended to add a comparative analysis with core existing studies to highlight the unique contributions of this research.
2. There is ambiguity in the interpretation of convergence results. The study states that African countries “follow similar trajectories but do not converge to the same steady state”, but does not explicitly explain whether this refers to intra-African convergence or convergence with global regions. It is suggested to clarify the scope of convergence in the discussion section to avoid reader confusion.
3. The machine learning model’s application lacks sufficient validation. The random forest model identifies fixed telephone lines as the key driver, but does not compare with other potential drivers or test model robustness through alternative algorithms. Adding these comparisons will enhance the credibility of the findings.
4. The policy recommendations need to be more actionable. The current suggestions of “deepening regional collaboration” and “balanced investment” are overly general. It is recommended to propose differentiated strategies based on the convergence characteristics of different technologies and cite specific regional cooperation cases for reference.
5. The systematic digitalisation framework lacks empirical verification. It is suggested to add regression analysis or case studies to verify the model’s applicability in Africa’s digitalisation process.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

This manuscript combines the Phillips–Sul club convergence approach with ML techniques to analyse Africa’s digitalisation trajectory. The approach is original since it combines convergence econometrics with predictive ML models (Random Forest and XGBoost). This approach is  uncommon in ICT convergence studies. Using a panel of 123 countries (1990–2023), evidence of intra-regional convergence in digitalization, in internet access and mobile cellular adoption is found. African economies have similar digital development trajectories, The digital diffusion gap gets narrow relative to other global regions, although they do not converge to the same steady state as advanced economies.

Major revisions:

1.There are inconsistencies between the reported empirical results and the interpretation in the discussion and conclusion sections. In Table 4, the study reports failure to reject convergence for the global sample and Africa. In conclusions, it states that the convergence hypothesis was rejected for the full panel and African region. This contradiction must be corrected.

Clarify the distinction between convergence within clubs, and convergence toward a global steady state.

Explain the interpretation of β and σ (speed of adjustment). Positive β values signify “strong convergence,” but the economic meaning of the magnitude is not sufficiently discussed.

2. The digitalisation process framework is descriptive. Connect it to the diffusion of innovation theory or technology adoption models. Explain the advantages of this study beyond the ICT literature (e.g., Park et al., Saba & David, Rath).

Justify better the combination of Phillips–Sul club convergence with ML forecasting.

3. ML methodology is not sufficiently developed. Give the following details:

- the prediction target? (convergence gap? DIGIX? Growth rate?)

-How was the dataset split (train/test)?

- if cross-validation used was used.

-How were hyperparameters selected?

- What performance metrics were used (RMSE, MAE, R2)?

-How does ML forecasting improve upon traditional econometric forecasting?

4. Justify why only PC1 was retained. Explain in detail the standardization procedures. Expand the sensitivity analysis using other weighting schemes. Justify how missing data were interpolated using excel trend formulas.
5. The convergence tests show divergence in fixed telephone lines. Random Forest identifies fixed telephone infrastructure as the main driver of Africa’s digital diffusion. These two findings appear contradictory.
6. Policy implications should be more tightly connected to the empirical findings. Strengthen empirically the recommendation for balanced reinvestment in fixed-line infrastructure. Argue the link between convergence clubs and regional digital policy coordination.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

It has solved my concerns

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

The economic meaning of the magnitude for alpha and beta could still be clearer. Explain what alpha = 2.56 implies in practical terms. 

In Table 4, the country list for “Asia and Pacific” is identical to the list for “Arab region”.

Author Response

Please see the attachment

Author Response File: Author Response.docx