Do Free Trade Agreements Facilitate FDI Spillover Effects on Domestic Firms? Empirical Evidence from Oman
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThank you for the manuscript. My comments are below.
The definition of population, sample size, and sampling method are not defined in the paper. Authors wrote that “A Kruskal-Wallis H-test test (one-way ANOVA)” . However, Kruskal Wallis cannot be defined as non-parametric ANOVA. It is used if the assumptions of ANOVA cannot be held. In Kruskal Wallis, we test the equality of the median but the equality of the mean in ANOVA. This confusion shows the authors' lack of knowledge of the subject. Moreover, in Table 4 it was written “dependent variable”. What are independent variables, they are not any, because in ANOVA and Kruskal Wallis, there is no dependent or independent variable. It is not used to see the relationship between two or more variables. The study is important because it is a case study and has originality. Without this originality, it would have been correct to reject it on these lines in practice. The authors have not mastered their practice. I hope they will make a careful correction. I wish them success.
Author Response
Does the introduction provide sufficientbackground and include all relevant references? (can be improved)
Response 1: The introduction has been improved with additional references highlighted in yellow. We also added more aspects that are relevant to the topic. For instance, FTAs strenthen economic ties between member states of the agreements (Cao et al., 2023). International trade significantly improves the productivity and technological level of domestic enterprises (Pietrucha and Zelazny, 2020; Angelini and Generale, 2008). Benmamoun et al. (2016) undertook a study focusing on the impact of international trade in terms of FDI, on 18 Arab countries for the period 1990 to 2011. They observed acceleration to the economic and human development.
- Is the research design appropriate? (can be improved)
Response 2: Thanks for your suggestion. An improvement in the research has been incorporated,
For instance, The data was taken from the population via random sampling. In this study, the following steps were used: we created a list of 600 companies operating in different parts of Oman and assigned them unique identifiers. Following it we brainstormed and con-cluded to get 457 companies in our sample. Then we randomly selected name of 457 companies from 600 and contacted them for the survey distribution. Thus it helped in minimizing the bias and allowed to confidently generalize the findings from the sample.
- Are the methods adequately described? (must be improved)
Response 3: The research methods further explained with addition of two subsections namley sampling method, Kruskal Wallis H test, higlighted in yellow.
- Are the results clearly presented? (must be improved)
Response 4: Thanks for your feedback. More explanations are provided in the results and discussion higlighted in yellow.
- Are the conclusions supported by the results? (can be improved)
Response 5: Additional information are added to the conclusion. For instance, the companies with 51% foreign stakes reported a higher share of knowledge transfer and product efficiency. The 75% foreign stakes reported higher impact of FDI in job creation and capital investment. Analysis also reveals that the Omani companies are interested more in FDI that facilitates technology and knowledge transfer and capital investment than focusing on developing R&D capabilities. The FDI priority areas in Oman are high as technology transfer (29%), knowledge transfer (18%), corporate management (18%), and capital investment (17%) to technical know-how (10%) and corporate culture (7%).
Suggestion
The definition of population, sample size, and sampling method are not defined in the paper.
The definition of population, sample size, and sampling method are defined with addition of subsections sampling method.
Sampling method and population
The survey target group are the companies located in Oman. The study uses random sampling methods. There are two categories of sampling methods: probability and non-probability. In the probability sampling, the samples are randomly selected from the population, while non-probability sampling is based on the researcher judgment and convenience (Copas and Li, 1997). This could lead to bias. In a random sampling technique, each member of the population gets an equal and independent chance of being chosen for a sample (Etikan and Bala, 2017; Singh and Masuku, 2014). Therefore, no association between the observations in each group or between the groups themselves, i.e. each individual or object in the population is chosen completely at random, with no bias in the selection process. This means that each individual or object in the population is chosen completely at random, with no bias in the selection process.
The data was taken from the population via random sampling. In this study, the following steps were used: we created a list of 600 companies operating in different parts of Oman and assigned them unique identifiers. Then, we brainstormed and concluded to get 457 companies in our sample. Then we randomly selected names of 457 companies from 600 and contacted them for the survey distribution. This helped in minimizing the bias and allowed to confidently and generalize the findings from the sample.
Authors wrote that “A Kruskal-Wallis H-test test (one-way ANOVA)” . However, Kruskal Wallis cannot be defined as non-parametric ANOVA. It is used if the assumptions of ANOVA cannot be held. In Kruskal Wallis, we test the equality of the median but the equality of the mean in ANOVA. This confusion shows the authors' lack of knowledge of the subject. Moreover, in Table 4 it was written “dependent variable”. What are independent variables, they are not any, because in ANOVA and Kruskal Wallis, there is no dependent or independent variable. It is not used to see the relationship between two or more variables. The study is important because it is a case study and has originality. Without this originality, it would have been correct to reject it on these lines in practice. The authors have not mastered their practice. I hope they will make a careful correction. I wish them success.
The subsection Kruskal Wallis H test provides explanation of Kruskal Wallis cannot be used as non-parametric with refernces.
The Kruskal-Wallis H test (sometimes referred to as the "one-way ANOVA on ranks") is a non-parametric test that uses ranks to determine if there are statistically significant differences between two or more groups of an independent variable on a continuous or ordinal dependent variable (Kruskal, 1952).
There are different levels in table 4 namley percentage of ownerships in foreign company, public company, local private company, and no partnership.
Revised text:
Kruskal Wallis H- tests
The study used Kruskal-Wallis H test. The Kruskal-Wallis H test is a rank-based nonparametric test that can be used to evaluate if two or more groups of an independent variable have statistically significant differences on a continuous or ordinal dependent variable if data is non-normal (McKight and Najab, 2010; Anderson, 2001; Siegel, 1957). This means to compare two or more independent samples with similar or different sample sizes.
It is regarded as a nonparametric alternative to one-way ANOVA (Kruskal, 1952) , as well as an extension of the Mann-Whitney U test (only used to compare two groups) to comparing k independent samples or the comparison of multiple independent groups (Miller, 1997; Breslow, 1970). Non-parametric means that it makes no assumptions about the data's parameters, such as mean, variance, and so on. This implies at least one sample is stochastically dominant over the other.
The Kruskal-Wallis H test is applicable under the conditions: 1) sampling or observations should be independent, 2) data normality (data in each category should be regularly distributed) should not be valid, 3) the dependent variable should be either the ordinal or interval or ratio, implying that they have some form of hierarchy, 4) the independent variable should have two or more categorized independent groups. A Kruskal-Wallis test requires several independent random samples with at least ordinally scaled features. Ordinal variables are sufficient in the Kruskal-Wallis H test since non-parametric tests use ranks rather than value differences.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe paper suggests rich empirical evidence generated by a large sample survey of Oman located companies regarding various issues of FDI spillovers. The content should be of particular interest to researchers studying this phenomenon especially in the regional context of front Asia, Arabic peninsula, etc.
In respect of the research design, the sample selection does not become clear. Expression like "A sample of 457 responses were collected from different cities across Oman...", "The survey covered companies located in...", "The companies approached are..." seems not enough - how the sample has been selected? Was a statistical office company register used? Or a list of companies registered at a business association, Yellow pages, etc.? Something more concrete about the sample frame and selection method is necessary to be clarified.
I would recommend some revisions of the text concerning the interpretation of results. For example,
* the sentence on lines 400-403:
"A sample of 457 responses were collected... After cleaning the data, 438 samples were analyzed."
Plural should not be used here, I would suggest reformulation e.g.
"After cleaning the data, 438 cases (or observations) were analyzed."
* the sentence on lines 556-558:
"No significant 556 difference is observed between technology transfer and medium and large size companies 557 (W= -2.91, p= .1), and small and medium size companies (W= -1.5, p= .537)."
There cannot be a difference between "technology transfer" and "companies" (small and medium size), but there is a difference "between small and medium size companies regarding the technology transfer".
No
Author Response
Point 1: Is the research design appropriate? (can be improved)
Response 1: Thanks for your suggestion. An improvement in the research has been incportaed,
For instance, The data was taken from the population via random sampling. In this study fol-lowing steps were used: we created a list of 600 companies operating in different parts of Oman and assigned them unique identifiers. Following it we brainstormed and con-cluded to get 457 companies in our sample. Then we randomly selected name of 457 companies from 600 and contacted them for the survey distribution. Thus it helped in minimizing the bias and allowed to confidently generalize the findings from the sample.
Point 2: Are the methods adequately described? (can be improved)
Response 2: The research methods further explained with addition of two subsections namley sampling method, Kruskal Wallis H test.
Sampling method and population
The survey target group are the companies located in Oman. The study uses random sampling methods. There are two categories of sampling methods: probability and non-probability. In the probability sampling, the samples are randomly selected from the population, while non-probability sampling is based on the researcher judgment and convenience (Copas and Li, 1997). This could lead to bias. In a random sampling technique, each member of the population gets an equal and independent chance of being chosen for a sample (Etikan and Bala, 2017; Singh and Masuku, 2014). Therefore, no association between the observations in each group or between the groups themselves, i.e. each individual or object in the population is chosen completely at random, with no bias in the selection process. This means that each individual or object in the population is chosen completely at random, with no bias in the selection process.
The data was taken from the population via random sampling. In this study, the following steps were used: we created a list of 600 companies operating in different parts of Oman and assigned them unique identifiers. Then, we brainstormed and concluded to get 457 companies in our sample. Then we randomly selected names of 457 companies from 600 and contacted them for the survey distribution. This helped in minimizing the bias and allowed to confidently and generalize the findings from the sample.
Kruskal Wallis H- tests
The study used Kruskal-Wallis H test. The Kruskal-Wallis H test is a rank-based nonparametric test that can be used to evaluate if two or more groups of an independent variable have statistically significant differences on a continuous or ordinal dependent variable if data is non-normal (McKight and Najab, 2010; Anderson, 2001; Siegel, 1957). This means to compare two or more independent samples with similar or different sample sizes.
It is regarded as a nonparametric alternative to one-way ANOVA (Kruskal, 1952) , as well as an extension of the Mann-Whitney U test (only used to compare two groups) to comparing k independent samples or the comparison of multiple independent groups (Miller, 1997; Breslow, 1970). Non-parametric means that it makes no assumptions about the data's parameters, such as mean, variance, and so on. This implies at least one sample is stochastically dominant over the other.
The Kruskal-Wallis H test is applicable under the conditions: 1) sampling or observations should be independent, 2) data normality (data in each category should be regularly distributed) should not be valid, 3) the dependent variable should be either the ordinal or interval or ratio, implying that they have some form of hierarchy, 4) the independent variable should have two or more categorized independent groups. A Kruskal-Wallis test requires several independent random samples with at least ordinally scaled features. Ordinal variables are sufficient in the Kruskal-Wallis H test since non-parametric tests use ranks rather than value differences.
Suggestions for Authors
- how the sample has been selected? Was a statistical office company register used? Or a list of companies registered at a business association, Yellow pages, etc.? Something more concrete about the sample frame and selection method is necessary to be clarified.
Response 3: Thanks for your comment. The conttents are modified with detailed explanation of sampling method and population.
Sampling method and population
The survey target group are the companies located in Oman. The study uses random sampling methods. There are two categories of sampling methods: probability and non-probability. In the probability sampling, the samples are randomly selected from the population, while non-probability sampling is based on the researcher judgment and convenience (Copas and Li, 1997). This could lead to bias. In a random sampling technique, each member of the population gets an equal and independent chance of being chosen for a sample (Etikan and Bala, 2017; Singh and Masuku, 2014). Therefore, no association between the observations in each group or between the groups themselves, i.e. each individual or object in the population is chosen completely at random, with no bias in the selection process. This means that each individual or object in the population is chosen completely at random, with no bias in the selection process.
The data was taken from the population via random sampling. In this study, the following steps were used: we created a list of 600 companies operating in different parts of Oman and assigned them unique identifiers. Then, we brainstormed and concluded to get 457 companies in our sample. Then we randomly selected names of 457 companies from 600 and contacted them for the survey distribution. This helped in minimizing the bias and allowed to confidently and generalize the findings from the sample.
- I would recommend some revisions of the text concerning the interpretation of results. For example,
* the sentence on lines 400-403:
"A sample of 457 responses were collected... After cleaning the data, 438 samples were analyzed."
Plural should not be used here, I would suggest reformulation e.g.
"After cleaning the data, 438 cases (or observations) were analyzed."
Response 4: Thanks for your suggestion. The texts are corrected.
A sample of 457 responses were collected from different cities across Oman, using both online and offline survey questionnaires, between 1 August 2023 to 31 October 2023. After cleaning the data, 438 cases (or observations) were analyzed.
- the sentence on lines 556-558:
"No significant 556 difference is observed between technology transfer and medium and large size companies 557 (W= -2.91, p= .1), and small and medium size companies (W= -1.5, p= .537)."
There cannot be a difference between "technology transfer" and "companies" (small and medium size), but there is a difference "between small and medium size companies regarding the technology transfer".
Response 5: Thanks for your suggestions, the sentences have been shortend and restructured to make it more meaningful.
No significant difference is observed between technology transfer and companies (medium and large size companies) (W= -2.91, p= .1), but there is a difference "between small and medium size companies regarding the technology transfer (W= -1.5, p= .537). These findings suggest that medium and small companies have the same spillover effect of FTA in terms of technology transfer.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsThis paper is carefully written and is a useful contribution to the literature. In a revision, i would like the authors to address the issue of why trade agreements are better than unilateral free trade. One paragraph should be enough.
Author Response
Point 1: why trade agreements are better than unilateral free trade
Response 1: Please find explanation of why free trade agreements are better than unilateral free trade in the paragraph below.
Free trade agreements offer a structured framework for countries whose primary objective is to remove trade obstacles such as tariffs and non-tariff barriers, promoting reciprocal benefits for participating countries (Freund and Ornelas, 2010; Menon, 2007). Trade agreements can be unilateral, bilateral, regional, or multinational. While unilateral trade agreements are one-sided, non-reciprocal trade privileges offered by a developed country to developing countries, however, they lack elements of assurance and reciprocity, potentially leaving a country vulnerable to exploitation or sudden shifts in global trade dynamics (Conway, 1989). Meanwhile, free trade agreements that involve more than two member states are better than unilateral free trade because they offer wider market access for companies operating in the regional market. FTAs also offer a wide range of incentives for member states, including national treatment, free movement of goods and services in the regional markets, free movement of factors of productions, which reduce the cost of production and make products more competitive compared to their counterparts in the non-member states. For some countries, unilateral measures are the most effective strategy to lower domestic trade obstacles. Other countries prefer FTAs, which have two advantages over unilateral approaches (Irwin, 2020). First, when many countries or regions agree to reduce trade barriers mutually, the economic benefits of international trade are strengthened and enhanced. FTAs among countries or regions are an effective tool for liberalizing regional and global trade (Bhagwati, 2002). Regional and multilateral reductions in trade barriers may lessen political opposition to free trade in each of the participating countries.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThank you for your corrections.
