Analysis of Exogenous Factors to Thailand Coffee Price Volatility: Using Multiple Exogenous Bayesian GARCH-X Model

Round 1

Reviewer 1 Report

Referee report on “Analysis of exogenous factors to Thailand coffee price volatility: using multiple exogenous Bayesian GARCH-X model”

This paper uses annual data for the period 1999-2021 to estimate the determinants of coffee price volatility in Thailand using a Multiple exogenous Bayesian GARCH-X model. Four determinants of coffee price volatility are examined: demand for coffee beans in Thailand, coffee bean exports in Indonesia, exports in Brazil and the volume of Vietnam’s coffee exports.  The authors find that domestic demand for coffee beans and the volume of Indonesia’s coffee beans exports affect positively the volatility of Thailand’s coffee price whereas the volume of Brazil’s export volume affects the volatility in Thailand’s coffee price negatively.

Comments:

(1)      The empirical analysis suffers from a major drawback which is the use of a very small sample size (only 23 annual observations) to perform GARCH analysis.  It is well known that GARCH estimation assumes a very large sample size which usually includes high-frequency data necessary to capture the feature of volatility clustering.  Volatility clustering is rarely found in annual data and in such cases the time series under consideration covers a long time period.  I, therefore, suggest the use of monthly data for the time period under consideration or a longer period, if available.  I addition, I suggest testing for GARCH effects using the well-known Engle’s approach where squared OLS residuals are regressed on their lags.  This test should be performed prior to the adoption of the GARCH approach.

(2)      The paper provides some summary statistics in table 2 which to an extent indicate the likely absence of GARCH effects in some variables. In particular, for variables TFP and TDD the probability value of the Jarque-Bera test is very high, thus not rejecting the null hypothesis of normality. This finding does not square with one of the features of GARCH effects which is deviation from normality.

(3)      The introductory section needs to be enriched with some features that are usually included in this section.  First, I suggest including the main results obtained in this study. Second, at the end of this section, I suggest you include a paragraph that outlines the following sections of the paper.  This is standard procedure in published manuscripts.

(4)       I suggest you explain better the statistical significance of the estimated coefficients reported in Table 5. It seems to me that some coefficients are not statistically significant and the constructed confidence intervals are incorrect.  For example, the alpha coefficients in models (2)-(5) seem to be statistically insignificant as the estimated coefficients are not at least equal to twice their standard errors. The same seems to be the case for the beta coefficients in models (1), (2), (4) and (5).

(5)      I suggest you include a subsection where the results obtained are placed in context of the related literature. This subsection could be included in section 4.

Other comments:

(1)    There should be a one-to-one correspondence between text citations and the reference list.  There are several papers cited in the text but not included in the reference list. For examples, the papers by Bollerslev (1986), Andersen et al (2003), Hansen and Lund (2004), etc.

(2)   Page 6: The parameters appearing in equation (7) are ω, α and β and not α₀, α₁ and β, as stated three lines below equation (7).

(3)   Table 5: In the notes, standard deviations are said to be include in parentheses. However, no figures appear inside parentheses. The figures in square brackets are confidence intervals.

I suggest the manuscript be read by an English-language editor.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear authors,

The abstract is too long and should be restructured so that it contains the following elements: purpose, design/methodology/approach, findings, originality, research limitations/implications, practical implications, social implications.

The way bibliographic sources are cited in the text does not meet the requirements of the journal to which the article was submitted. Authors should read the editorial guidelines carefully.

The introduction does not reveal the research problem, nor the gap identified in the literature on the researched problem. The authors should review these issues. The description of the model used, with quotations from the literature, has no place in the introductory part. At the end of the introduction, the authors should present how the paper is structured, by sections. In conclusion, the introductory part should be rewritten and almost entirely rewritten.

The literature review is also not well written. It should refer not only to previous studies on the model used in the methodology, but primarily to international studies on the coffee market and the exogenous factors affecting its price volatility.

The methodology should contain both the material and the method used in the study and how the data were analysed. The material part is missing entirely. It is not specified what kind of data was analysed, where it came from, what period was analysed, etc. Is it secondary or primary data? This section of the article should also be completely rewritten, as it does not meet the requirements of a scientific article.

The results are well presented but not discussed at all. They should be compared with the results of previous studies. Therefore it is important to identify these studies in the literature review part.

Conclusions should be reformulated. Much of the text repeats what has already been written in previous sections. In the conclusions part, we should not have quotations from the literature. They should highlight their own results.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf