Financial Literacy as a Driver of Financial Inclusion in Kenya and Tanzania

Round 1

Reviewer 1 Report

• P1-3 Introduction: I missed information on the contribution of your paper in relation with the existing literature. What’s actually new in contrast to what we already know from previous literature.
• P3 Data: What is exactly an enumerator area. It is difficult to grasp this concept? How big is this? How is this division made?
• P3: There seems something wrong with this sentence: “Households within each EA are used as secondary sampling units and respondents are randomly drawn from interview from qualifying household members.”
• P3: “The sample data are weighted back at provincial level to the population”. What does that mean? Is your sample representative for the population in each country?
• P4: It is not always clear what you understand under financial literacy and what under financial behaviours? ‘tracking spending and income’ is this financial literacy or financial behaviour? In one of your econometric models you are planning to link financial literacy with financial behaviours. So it must be clear from the start what you mean exactly. Actually, do you mean really financial behaviours or is it more a link between financial literacy and financial inclusion?
• P4: You refer to Appendix A, but I do not find an appendix.
• P5: Table 2. Make clear immediately that it are percentages that are shown. Mentioning ‘proportion’ is not immediately clear.
• P6: you refer to Table 2 when you talk about ‘credit penetration’ but I do not find any information on credit penetration in your table.
• P6-7: I would suggest to replace ‘sex’ by ‘gender’. The latter is more common in the financial literacy literature.
• P7: “As shown in Table 3, there is a slight Sex difference with 56 percent of women tracking spending and earnings compared to 59 percent of men.” I do not find this information in Table 3. Table 3 only shows some info on ‘understanding of a bank account’. It seems that the other aspects discussed previously are missing in this table. Please check Table 3 carefully as a lot of info is missing.
• Table 3: please use the same order for the different parameters in the table as the discussion on P7 and 8. Now, it is confusing for the reader.
• TABLE 4??? I do not see a Table 4. We go from Table 3 to Table 5. Please renumber if there is no Table 4 missing.
• P9: What is an ‘informal bank account’?
• P10: What is ‘informal saving’?
• P11: in Table 6 you mention ‘sources of financial advice’ without specifying what are exactly the different sources.

Author Response

We are grateful for constructive comments. Our paper immensly benefited from the feedback. 

• P1-3 Introduction: I missed information on the contribution of your paper in relation with the existing literature. What’s actually new in contrast to what we already know from previous literature.

 

The paper represents the first attempt to examine the link between financial literacy and financial inclusion in the context of developing economies in Africa. This has been explained in the first section.

• P3 Data: What is exactly an enumerator area. It is difficult to grasp this concept? How big is this? How is this division made?

Enumerator areas comprise hundreds of households in a given district. In a multi-stage sampling a country is divided into districts and districts into enumerator areas. Finally households are randomly selected form each enumerator area.

 

• P3: There seems something wrong with this sentence: “Households within each EA are used as secondary sampling units and respondents are randomly drawn from interviewfrom qualifying household members.”

The sentence has been revised.

 

• P3: “The sample data are weighted back at provincial level to the population”. What does that mean? Is your sample representative for the population in each country?

Yes the sample is nationally representative

 

• P4: It is not always clear what you understand under financial literacy and what under financial behaviours? ‘tracking spending and income’ is this financial literacy or financial behaviour? In one of your econometric models you are planning to link financial literacy with financial behaviours. So it must be clear from the start what you mean exactly. Actually, do you mean really financial behaviours or is it more a link between financial literacy and financial inclusion?

The initial dataset was abandoned and we used a new dataset that included questions on financial literacy following OECD toolkit

 

• P4: You refer to Appendix A, but I do not find an appendix.

This has been corrected

 

• P5: Table 2. Make clear immediately that it are percentages that are shown. Mentioning ‘proportion’ is not immediately clear.

This has been removed

 

• P6: you refer to Table 2 when you talk about ‘credit penetration’ but I do not find any information on credit penetration in your table.

The text and table are no longer in the revised paper

 

• P6-7: I would suggest to replace ‘sex’ by ‘gender’. The latter is more common in the financial literacy literature.

 

We did replace sex by gender

• P7: “As shown in Table 3, there is a slight Sex difference with 56 percent of women tracking spending and earnings compared to 59 percent of men.” I do not find this information in Table 3. Table 3 only shows some info on ‘understanding of a bank account’. It seems that the other aspects discussed previously are missing in this table. Please check Table 3 carefully as a lot of info is missing.

This has been removed and new tables have been populated

 

• Table 3: please use the same order for the different parameters in the table as the discussion on P7 and 8. Now, it is confusing for the reader.

This has been addressed in the revised paper

 

• TABLE 4??? I do not see a Table 4. We go from Table 3 to Table 5. Please renumber if there is no Table 4 missing.

All tables are correctly referenced in the revised manuscript

 

• P9: What is an ‘informal bank account’?

This is no longer there

 

• P10: What is ‘informal saving’?

This is no longer there

 

• P11: in Table 6 you mention ‘sources of financial advice’ without specifying what are exactly the different sources.

This is no longer there

Reviewer 2 Report

The manuscript meets the expectations for publication.

This paper examines the role of financial literacy in encouraging households to participate in financial markets and use financial products.

Data from 11 Southern African countries were used. The data were from FinScope Surveys.

Logistic regression was used to explain 6 variables (Tables 5 and 6). This is explained in section 2.2. The 6 variables are assorted measures of financial market/product use.

Comments/suggestions:

After section 2.2 one would expect to see results of the model discussed. But there are multiple pages of other material (sections 3.1.and 3.2 that would seem better put somewhere before 2.2). Then the results are shown in Section 3.3.

No descriptive statistics are shown for all the variables used.

The 6 variables could be explained more clearly or in a way where the reader knows better what the variables are in the top of Tables 5 and 6.

Tables 5 and 6 independent variables are not all the same. The first 2 is makes sense that they are different, but the remainder would seem to be the same but are not. Or maybe they are the same and were given different names in each Table? (Location is Urban/Rural?, Personal monthly income categories is Income?)

There is no Table 4 after 3 (there is a 5 and 6).

Why not use a dummy variable for each country (there are not many in the sample) and report that coefficient? This would replace the fixed effects as well.

There is a causality question. Do those using financial products obtain knowledge/literacy about them by using them? The set up here assumes that the causality goes from knowledge/literacy to use. There is no way simple to fix this bi-directional causality, but it can be mentioned that if there is the bi-directional causality the measurement of the impact of financial knowledge/literacy may be less than it appears.

Author Response

We are grateful for constructive comments. Our paper has immensly benefited from the feedback.

After section 2.2 one would expect to see results of the model discussed. But there are multiple pages of other material (sections 3.1.and 3.2 that would seem better put somewhere before 2.2). Then the results are shown in Section 3.3.

Results and discussions are presented together. The paper has taken a new shape after reducing the countries to two and using a more relevant data to measure financial literacy 

No descriptive statistics are shown for all the variables used.

Mean value of financial literacy are presented across socio-demographic variables used in the models

The 6 variables could be explained more clearly or in a way where the reader knows better what the variables are in the top of Tables 5 and 6.

The flow of the results section has changed significantly.

Tables 5 and 6 independent variables are not all the same. The first 2 is makes sense that they are different, but the remainder would seem to be the same but are not. Or maybe they are the same and were given different names in each Table? (Location is Urban/Rural?, Personal monthly income categories is Income?)

The tables are replaced by new ones

There is no Table 4 after 3 (there is a 5 and 6).

Error in labelling tables have been corrected

Why not use a dummy variable for each country (there are not many in the sample) and report that coefficient? This would replace the fixed effects as well.

We used data from only two countries and as a result the models are estimated for the two countries separately, eliminating the need to use country dummy to control country fixed effects.

There is a causality question. Do those using financial products obtain knowledge/literacy about them by using them? The set up here assumes that the causality goes from knowledge/literacy to use. There is no way simple to fix this bi-directional causality, but it can be mentioned that if there is the bi-directional causality the measurement of the impact of financial knowledge/literacy may be less than it appears.

We estimated instrumental variable model to address endogeniety often caused by reverse causality, omitted variable bias and measurement error.

Reviewer 3 Report

The paper aims to link the financial literacy with the financial inclusion in some African countries. It shows that financial literacy are related to increased level of financial inclusion in these countries. 

While the paper provides us further evidence on the relationship between financial literacy and financial inclusion, several weak points needs to be addressed carefully. 

• Numerous papers have examined the relationship between financial literacy and financial inclusion in developing countries (for example Morgan and Trinh 2019; Morgan and Trinh 2020), the authors should clearly answer the question: what makes this study different from other studies?
• While there are various ways to measure the financial literacy, the paper should clearly explain their choice of financial literacy indicators and explain why such indicators are comparable to other financial literacy indicators widely used.
• Please clarify how to measure the financial knowledge by presenting which information from the survey is used for each indicator.
• The paper have very detailed information on financial literacy (in section 3) but gives very little attentions to the outcome variables: financial inclusion. In the end, what the paper wants to see is how financial literacy contributes to financial inclusion, not financial literacy per se.
• It is strange that the sample used for descriptive analysis included 11 countries while the regression analysis have only 6 countries. I think it is better to keep the sample consistent across analysis.
• The authors well recognize the reserve causality relations (and explain it in some places in the manuscript). This may lead to endogeneity issue. I guess the authors did not plan to derive the causal relationship between financial literacy and financial inclusion, but at least explain the endogeneity issues in their discussion.
• I think the authors should do various robustness with their results. I don't think the current version gives us enough confident to make conclusion.
• In the current form, the paper lacks a good discussion of the estimation results and how it link with previous literature.
• There may be heterogenous across countries, it is better to estimate for each country and to see how different each financial literacy indicator is associated with financial inclusion across countries. 

Referece: 

Morgan P. and Trinh Q.L., 2019. Determinants and Impacts of Financial Literacy in Cambodia and Viet Nam. Journal of Risk and Financial Management. Vol. 12, Art. 19; 

Morgan P. and Trinh Q,.L., 2020. Financial Literacy, Financial Inclusion, and Savings Behavior in Laos, Journal of Asian Economics. Vol. 68. Article No. 101197, 

Author Response

Thank you for your valuable feedback. Our paper immensly benefited from your insicive comments.

• Numerous papers have examined the relationship between financial literacy and financial inclusion in developing countries (for example Morgan and Trinh 2019; Morgan and Trinh 2020), the authors should clearly answer the question: what makes this study different from other studies?

 

Although there are similar studies in developing countries in Asia, none on Africa. Our paper brings the African context to the debate by focusing two East African economies where digital financial services are widespread.

 

• While there are various ways to measure the financial literacy, the paper should clearly explain their choice of financial literacy indicators and explain why such indicators are comparable to other financial literacy indicators widely used.

We abandoned FinScope datasets used in the original paper. The revised paper is based on financial inclusion insight database that captured two of the three dimensions of financial literacy as per OECD guideline.

 

• Please clarify how to measure the financial knowledge by presenting which information from the survey is used for each indicator.

 

Details on the measurement of financial knowledge and financial behaviour has been given in the appendix. The FII survey was designed following OECD guidelines.

 

• The paper have very detailed information on financial literacy (in section 3) but gives very little attentions to the outcome variables: financial inclusion. In the end, what the paper wants to see is how financial literacy contributes to financial inclusion, not financial literacy per se.

We discussed financial inclusion in the results section.

 

• It is strange that the sample used for descriptive analysis included 11 countries while the regression analysis have only 6 countries. I think it is better to keep the sample consistent across analysis.

 

Descriptive and regression analysis are now aligned because the countries in the dataset are only two.

 

• The authors well recognize the reserve causality relations (and explain it in some places in the manuscript). This may lead to endogeneity issue. I guess the authors did not plan to derive the causal relationship between financial literacy and financial inclusion, but at least explain the endogeneity issues in their discussion.

 

We really appreciate for this comment. To address endogeniety problem we estimated instrumental variable model. The results are now more robust that what was reported in the original paper.

• I think the authors should do various robustness with their results. I don't think the current version gives us enough confident to make conclusion.

We used instrumental variable regression to obtain more robust results.

• In the current form, the paper lacks a good discussion of the estimation results and how it link with previous literature.

We presented results and discussions together. Where necessary, we linked the findings with previous studies.

 

• There may be heterogenous across countries, it is better to estimate for each country and to see how different each financial literacy indicator is associated with financial inclusion across countries. 

We estimated OLS and IV regression for the two countries separately.

Reviewer 4 Report

The article is of low scientific value and its contribution to the knowledge of financial literacy is questionable.

First, the weakness of the article is its composition. The introduction has the character of a literature review. This in itself is not a bad solution and is accepted in some journals, but the introduction focuses too little on the research problem (what is it really?) and deals too much with the issue of defining financial literacy. The "Discussion" section has not been separated. From the content, it appears that the discussion is part of the "Results" section. This combination makes the article appear poorer.

Second, the "Method" section should provide full information about the variables used in the regression models: how they were measured in the survey and how they were coded for the analyses. It is also standard in articles of this nature to report descriptive statistics for the variables used, as well as at least brief information about the structure of the survey sample in terms of key socio-demographic variables. There is nothing of the sort in the article.

Third, perhaps the biggest drawback of the article is the conceptualization and operationalization of financial literacy adopted. The measure(s?) used are vague, not validated in any way; the article does not even include a robust discussion of how the measure is embedded in the existing literature.

Fourth, the article uses the language of causality to address the relationship between financial literacy and financial inclusion (which involves financial market participation-that is, consumer behavior). There is a rich literature that leaves no doubt that inferring the direction of the causal relationship between financial literacy requires either instrumental variables or longitudinal data.

Finally, the article does not add anything truly new and fresh to the knowledge of financial literacy. Its only strength is the use of a large (and unused in previous research) dataset. However, no use was made of it. Despite the fact that the article is based on data from Africa, the specifics of this area have not been presented and discussed in any way in the context of financial literacy.

Author Response

We are grateful for the insicive comments. Our  paper has immensly benefited from your valuable feedback.

The article is of low scientific value and its contribution to the knowledge of financial literacy is questionable.

Given that each economy has unique socio-economic realities, our paper that is based on the context of two lower middle income economies in Africa is an addition to a limited literature on the subject.

First, the weakness of the article is its composition. The introduction has the character of a literature review. This in itself is not a bad solution and is accepted in some journals, but the introduction focuses too little on the research problem (what is it really?) and deals too much with the issue of defining financial literacy. The "Discussion" section has not been separated. From the content, it appears that the discussion is part of the "Results" section. This combination makes the article appear poorer.

The introduction has been revised by removing the definition and addition a paragraph on the rationale of the study.

Second, the "Method" section should provide full information about the variables used in the regression models: how they were measured in the survey and how they were coded for the analyses. It is also standard in articles of this nature to report descriptive statistics for the variables used, as well as at least brief information about the structure of the survey sample in terms of key socio-demographic variables. There is nothing of the sort in the article.

Details on the two main variables, namely, financial literacy and financial inclusion are provided in the appendix.

Third, perhaps the biggest drawback of the article is the conceptualization and operationalization of financial literacy adopted. The measure(s?) used are vague, not validated in any way; the article does not even include a robust discussion of how the measure is embedded in the existing literature.

The revised manuscript follows OECD guideline on measuring financial literacy and financial inclusion.

Fourth, the article uses the language of causality to address the relationship between financial literacy and financial inclusion (which involves financial market participation-that is, consumer behavior). There is a rich literature that leaves no doubt that inferring the direction of the causal relationship between financial literacy requires either instrumental variables or longitudinal data.

We used instrumental variable model to estimate the link between financial literacy and financial inclusion.

Finally, the article does not add anything truly new and fresh to the knowledge of financial literacy. Its only strength is the use of a large (and unused in previous research) dataset. However, no use was made of it. Despite the fact that the article is based on data from Africa, the specifics of this area have not been presented and discussed in any way in the context of financial literacy.

There hasn’t been a similar study in the African context and hence we believe that our study serves are further validation of the theoretical link between financial literacy and financial inclusion. 

Round 2

Reviewer 1 Report

Thank you for your responses to my comments.

Author Response

Dear Sir,

We appreciate your valuable comments on the earlier version of the paper. The paper has been significantly improved based on your constructive comments.

Reviewer 3 Report

This paper seems to be a new paper rather than revising the previous one. It is fine to me since this version seems better. 

I specially focus on the estimations regarding the effects of financial literacy on financial inclusion. Authors report that they use IV approach, they need to explain their arguments for using the instrument and and how (and why) using instrumental variable change the estimation results if any. The first stage estimation results should also be presented. 

 

 

Author Response

Dear Sir,

Thank you for your constructive comments without which the paper could not have been brough to its current form. We are sepcially grateful for your comments on the methodology. Kindly see our responses to the latest version.

1. I specially focus on the estimations regarding the effects of financial literacy on financial inclusion. Authors report that they use IV approach, they need to explain their arguments for using the instrument and and how (and why) using instrumental variable change the estimation results if any.

      Response: we presented in few lines our explanation for the instruments we        used in the model.

      2. The first stage estimation results should also be presented. 

      We thought of presenting the first stage results as well but the table             becomes cumbersome with nine columns. However, we reported the F-stat from test of instruments where the null hypothesis that instruments are weak is rejected at 1%.