Financial Development and Language Structures
Round 1
Reviewer 1 Report
-
- The very first sentence of the manuscript (line 18-19) mentions that as per the literature, financial development has a positive effect on economic growth. This is not true. In fact, recent studies indicate a negative impact/non-significant impact of financial on economic growth. See Narayan and Narayan, 2013;Sy-Hoa et al. 2022. At best, one can say that the growth impact of financial development is mixed.
- The premise of the study is interesting. However, as shown by Roberts et al (2015), it is important to control for language family when weak/strong FTR languages are concerned. The current estimations do not control for this. See Roberts et al (2015): Future tense and economic decisions: controlling for cultural evolution PLoS One, 10 (7) (2015), pp. 1-46, 10.1371/journal.pone.0132145
- Although the author/s construct a panel, the main variable of interest (FTR) is time-invariant. Hence, I am not sure how having a panel estimation helps, more so since typical panel estimation (Fixed effects/Random effects) cannot be carried out in this context. Although the author/s control for time fixed effects, what about the country or continent fixed effects? I would suggest that a cross-sectional OLS be carried out, with required time and continent fixed effects. This would reduce the sample size, but this is mostly how these kinds of issues are dealt with in the literature. See Ang (2019)-
CULTURE, LEGAL ORIGINS, AND FINANCIAL DEVELOPMENT, ECONOMIC INQUIRY, 2019.
- Although the author/s rule out reverse causality, they agree that there might be omitted variable bias. Although as robustness checks more explanatory variables are introduced, the proper way to deal with this issue will be to use the Oster's method. Again, see the above reference (Ang (2019) for this.
- The authors rightly mention that the literature links time reference angle of FTR to time preference (discounting of the future). In this case, the authors mention that this is linked with savings etc. and that is affected by financial development. While intuitively this is correct, there is nothing in the model that shows how this inter-temporal aspect (due to FTR) manifest itself. See Dar and Sahu (2022) for the possible channel effects. Ref: The effect of language on financial inclusion. Economic Modelling. January 2022.
- Chen (2013) uses the verb ratio and the sentence ratio as alternate measures for weak-FTR. This should be done in the robustness check section for this work as well.
- Typos should be corrected (line 156- We use GDP per capita growth as a measure of financial development- it should be economic development; line 158- according to your hypothesis- should be our hypothesis; line 164- as a percentage of GDP- this has been mentioned twice.)
Author Response
Thanks for all your comments. Please see the responses in the attached file.
Author Response File: Author Response.pdf
Reviewer 2 Report
I appreciate the chance to read and review this interesting manuscript. While I think the paper makes an interesting point and am convinced the matter of grammatical influences on economic outcomes deserves continued attention, as a close follower of the related literature, I do have some comments and concerns about this study and the reliability of its main results.
1) While, as far as I know, this particular outcome has not been studied yet, there are several closely related papers that the authors should acknowledge. The review by Mavisakalyan & Weber (2018) lists those published up until 2018; there might be more in the meantime. With this in mind, the (verbal explanation of the) contribution of the paper might be sharpened.
2) The review itself should also be discussed in this context, as it points towards a number of challenges in (correctly) identifying such language effects that the authors should take into close consideration.
3) Related to the last point, there has been some criticism of the Chen 2013 AER paper from linguistics, particularly by Roberts & Winters (2013), resulting in a follow-up publication by Roberts, Winters & Chen (2015), where they further discuss these criticisms. The authors should at the very least discuss their findings in light of this debate.
3a) Roberts, Winters and Chen (2015) also suggest an empirical test to account for the inter-relatedness of language. The authors might want to consider applying this test to their empirical work (or check if it is applicable).
4) Also relatedly, I think it would help the argument of the paper to refer explicitly to the theoretical linguistic background, namely linguistic relativism / the Sapir-Whorf hypothesis. As it is, the underlying mechanism remains rather vague (line 132/133: "As Chen (2013) indicates, agents’ intertemporal preferences and decision making are represented via strong and weak FTR languages"). Also, credit where credit is due; economists did not come up with this idea.
5) Regarding the explication of the empirical strategy, I suggest presenting descriptive statistics of the relevant variables. In particular, the authors should clarify (early on, in Section 2.2) what unit the variable of interest is reported in and how it is distributed in their sample (how much variation is there to explain in the first place?).
6) Also related to the empirical strategy, I am missing an explicit explanation of how languages are assigned to countries. Is it the dominant language? What about multilingual countries? Maybe one could use a weighted (by share of speakers of individual languages) average in the latter case (at least as a robustness exercise)?
7) Related to the issues raised in Roberts & Winters (2013), Mavisakalyan & Weber (2018), and others, I think one cannot reasonably simply claim that the "main variable of interest is exogenous". If culture and languages co-evolve, and there is evidence to suggest it does, grammatical structures are not exogenous to behaviour. See also, for example, Beit-Hallahmi et al., 1974; Beller et al., 2015; Roberts and Winters, 2013; Galor et al., 2018.
A couple of smaller / less significant observations:
8) Female LFP is not mentioned in the list of covariates but still appears as one later in the paper.
9) There seems to be a typo in the note to Table 1.A (line 203): "and ) strong FTR countries"
10) It would improve the ad-hoc readability of the table if the main identifying variable was named "weak FTR" instead of just "FTR".
11) In the robustness section, on line 240, the authors claim that it is important to include "as many control variables as possible". I do not think that is true and I also don't think it is what the authors actually mean, but it would still help to more explicitly discuss the concerns about bias related to omitted variables at this point to make it more clear what it is the authors are trying to demonstrate, precisely.
12) A typo in line 286: seminar vs seminal
References:
Beit-Hallahmi, B. , Catford, J.C. , Cooley, R.E. , Dull, C.Y. , Guiora, A.Z. , Raluszny, M. , 1974. Grammatical gender and gender identity development: Cross cultural and cross lingual implications.. American Journal of Orthopsychiatry 44 (3), 424 .
Beller, S. , Brattebø, K.F. , Lavik, K.O. , Reigstad, R.D. , Bender, A. , 2015. Culture or language: what drives effects of grammatical gender? Cognitive Linguistics 26 (2), 331–359 .
Galor, O. , Özak, Ö. , Sarid, A. , 2018. Geographical Roots of the Coevolution of Cultural and Linguistic Traits. IZA Discussion Paper, 11982 .
Mavisakalyan, A., Weber, C., 2018. Linguistic structures and economic outcomes. Journal of Economic Surveys 32 (3), 916–939. doi: 10.1111/joes.12247 .
Roberts, S. , Winters, J. , 2013. Linguistic diversity and traffic accidents: Lessons from sta- tistical studies of cultural traits. PloS one 8 (8), e70902 .
Roberts, S.G. , Winters, J. , Chen, K. , 2015. Future tense and economic decisions: controlling for cultural evolution. PloS one 10 (7), e0132145 .
Comments for author File: Comments.pdf
Author Response
Thanks for all your comments. Please see the responses in the attached file.
Author Response File: Author Response.pdf
Reviewer 3 Report
Using cross-country data, this paper explores the effect of future time reference (FTR) on financial development. They found that countries that speak weak FTR languages (e.g. German and Chinese) have more enhanced financial development (measured by private credit to GDP) than the countries speaking strong FTR languages (e.g. French and English). The authors also showed that the results were robust to alternative measures of financial development. Overall, the study is a worthwhile extension of Chen's (2013) linguistic-savings hypothesis. I have two comments for the authors' consideration. First, perhaps in the discussion section, the authors should highlight that their findings are primarily correlational; readers should interpret the results with caution. Second, several recent studies (e.g., Chen, He, and Riyanto, 2019; Angerer et al., 2021; Jäggi et al., 2022) have experimentally examined the linguistic-savings hypothesis and did not find further support for the hypothesis. The authors should discuss these new findings either in the introduction section or in the discussion section to have a more balanced review of the literature.
Author Response
Thanks for all your comments. Please see the responses in the attached file.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
I appreciate the authors' effort to take my suggestions into consideration while revising the manuscript. I think the effort was worthwhile as the paper improved substantially.
I have no further comments, but this small one regarding language: There seems to be a minor error on page 7, last paragraph: the words "and other financial institutions" appear twice. And in the abstract, I think it should say "specifically", instead of "specially" and, later, "have enhanced" instead of "has". But these are just minor observations, that would probably also be picked up by the editorial team...