Thesis (Selection of subject)Thesis (Selection of subject)(version: 368)
Thesis details
   Login via CAS
The case of coal: A meta-analysis of demand and substitution
Thesis title in Czech: Případ uhlí: Meta-analýza poptávky a substituce
Thesis title in English: The case of coal: A meta-analysis of demand and substitution
Academic year of topic announcement: 2020/2021
Thesis type: Bachelor's thesis
Thesis language: angličtina
Department: Institute of Economic Studies (23-IES)
Supervisor: doc. PhDr. Zuzana Havránková, Ph.D.
Author: hidden - assigned by the advisor
Date of registration: 24.05.2021
Date of assignment: 24.05.2021
Date and time of defence: 08.06.2022 09:00
Venue of defence: Opletalova - Opletalova 26, O314, Opletalova - místn. č. 314
Date of electronic submission:02.05.2022
Date of proceeded defence: 08.06.2022
Opponents: Mgr. Josef Bajzík
URKUND check:
Amini, S. M. & C. F. Parmeter (2012): Comparison of model averaging techniques: Assessing growth determinants. Journal of Applied Econometrics 27(5), 870-876.
Andrews, I. & M. Kasy (2019): Identification of and Correction for Publication Bias. American Economic Review 109(8), 2766–2794.
Apostolakis, B. E. (1990): Interfuel and energy-capital complementarity in manufacturing industries. Applied Energy 35(2), 83-107.
Bacon, R. (1992) Measuring the possibilities of interfuel substitution. Policy Research Working Paper Series 1031. The World Bank: Washington, DC.
Bom, P. R. D. & H. Rachinger (2019): A Kinked Meta-Regression Model for Publication Bias Correction. Research Synthesis Methods 10(4), 497-514.
Chen, S. (2017): Interfuel Substitution in Electricity Generation Sector: A Meta-analysis. Master’s thesis, Department of Economics of the University of Ottawa, 27 pages.
Cohen A. J., Anderson H. R., Ostro B., Pandey K. D., Krzyzanowski M., & K. Net (2004): Urban air pollution. In: Comaparative Quantification of Health Risks (eds.: Ezzati M., Rodgers A., Lopez A., & C. Murry), volume 2, Geneva: World Health Organization, 1353-1433.
Considine, T. J. (1989): Separability, functional form and regulatory policy in models of interfuel substitution. Energy Economics 11(2), 82-94.
Egger, M., G. D. Smith, M. Schneider, & C. Minder (1997): Bias in meta-analysis detected by a simple, graphical test. British Medical Journal 315(7109), 629-634.
Furukawa, C. (2019): Publication Bias under Aggregation Frictions: Theory, Evidence, and a New Correction Method. Unpublished paper, MIT.
Hedges, L. V. (1992): Modeling publication selection effects in meta-analysis. Statistical Science 7(2), 246-255.
IEA (2020): CO2 Emissions from Fuel Combustion. Online at (accessed on April 29th, 2021).
Ioannidis, J. P., T. D. Stanley, & H. Doucouliagos (2017): The Power of Bias in Economics Research. Economic Journal 127(605), 236-265.
Jadidzadeh, A. & Serletis, A. (2016). Sectoral Interfuel Substitution in Canada: An Application of NQ Flexible Functional Forms. The Energy Journal, 37(2), 181-199
Renou-Maissant, P. (1999): Interfuel competition in the industrial sector of seven OECD countries. Energy Policy 27(2), 99-110.
Serletis, A., Timilsina, G. R., & Vasetsky, O. (2009): On Interfuel Substitution: Some International Evidence. Policy Research Working Paper 5026, World Bank.
Smyth, R., Narayan, P. K., & Shi, H. (2012): Inter-fuel substitution in the Chinese iron and steel sector. International Journal of Production Economics 139(2), 525-532.
Stanley, T. D. (2005): Beyond Publication Bias. Journal of Economic Surveys 19(3), 309-345.
Stern, D. I. (2012): Interfuel substitution: A meta-analysis. Journal of economic surveys 26(2), 307-331.
van Aert, R. C. & M. van A. (2021) Correcting for Publication Bias in a Meta-Analysis with the P-uniform* Method. doi: 10.3969/j.issn.1009-4393.2017.34.029.
Preliminary scope of work in English
Research question and motivation

The consumption of coal is a major source of greenhouse gas emissions destabilizing the climate (IEA, 2020). It is also the cause of air pollution linked to respiratory, cardiovascular, and neurological diseases (Cohen et al., 2004). The more the industries are willing to substitute for cleaner energy sources, the lower the cost of environmental policy that mitigates the unwanted externalities of coal consumption will be. This willingness of substitution, or the inter-fuel elasticity of substitution is, therefore, a key input of many environmental models that investigate the impacts of such policies. But how large are these elasticities?

The seminal literature surveys of Apostolakis (1990) and Bacon (1992) already document a large variation in the magnitude of the interfuel elasticities. For instance, some studies find coal and gas to be complements (Considine, 1989) while others state that they are substitutes (Smyth et al., 2012). The values are reported to vary across countries (Renou-Maissant, 1999), as well as across different industries (Serletis et al., 2009) and sectors (Jadidzadeh & Serletis, 2016). The first attempt to analyze this variation quantitatively can be found in Stern (2012). This carefully conducted meta-analysis explored the manufacturing-industry and country-level estimates and was closely followed by Chen (2017), who focused on the electricity generation sector. Both of them look for the correlates that could explain the differences in magnitudes. Both of them, however, missed several important features that should be addressed by any modern rigorous meta-study: publication bias and model uncertainty. In my thesis, I want to build on their work and dedicate special attention to coal, the largest pollutant of all.


In my thesis, I plan to pick up on the previous meta-analyses but control for several unaddressed features. Naturally, these studies have older datasets that do not account for the most recent developments in energy usage: many countries committed since 2015 to increase their shares of renewables and recent primary studies would account for these developments. Therefore, I will update the previous datasets with newer studies but also check for the studies the meta-analysts could potentially miss. Second, neither of these meta-analyses addressed the issue of publication selection bias rigorously. I will use state-of-the-art tools to analyze and filter out the bias (if any). Furthermore, I will show why the elasticities differ using methodologies that treat for the model uncertainty, the model averaging techniques. Finally, I plan to provide an in-depth discussion of the drivers of the effect, which is fairly brief in both previous meta-analyses, and estimate the best-practice effects consequently translated to a monetary value.


My intention is to focus on coal (the pairwise elasticities related to the substitution between coal and other fuel) and address the effects related to residential, industrial, and electricity-generation sectors of the economies, and possibly the macro-effects as well. If the collection proves to be too laborious (there could be thousands of estimates), I will refocus my search to sectoral values only. To collect the new studies and estimates I will use Google Scholar, which is superior to other search engines, so as not to exclude any recent and relevant information.

Besides collecting the substitution elasticities, I will also collect their measures of error to test for the presence of publication bias. First, I will construct a so-called funnel plot (Egger et al.,1997) which is a scatter plot with estimated effect on the horizontal axis and measure of error on the vertical axis. In case there is no publication bias present, the scatter plot should be symmetrical and hollow. The funnel plot is, however, only a visual test and the conclusions based on such test are fairly subjective. Therefore, I will perform a more rigorous Funnel Asymmetry test proposed by Stanley (2005) which analyzes the existence of the correlation between the estimate and its standard error. Because the relationship between the publication bias and the standard error does not necessarily have to be linear, several non-linear tests will follow, such as Top10 (Stanley, 2010), the selection model based on Hedges (1992) by Andrews & Kasy (2019), the weighted average of adequately powered by Ioannidis et al. (2017), the stem-based method by Furukawa (2019) or the endogenous kink-model by Bom & Rachinger (2019). At last, I will perform p-uniform* test (van Aert & van Assen, 2021) from psychology which does not assume any relationship between the estimate and its standard error and instead, uses p-values to identify publication bias.

Eventually, I will analyze the remaining heterogeneity in the literature. This includes coding many explanatory variables that will describe the context in which the studies are estimated. Indeed, it is not apparent beforehand which of these variables are important in our model (because there is no exact theory that would tell us). To account for model uncertainty, I will apply the Bayesian model averaging as a baseline model. To make sure my results are robust under different model specifications, I will tweak the priors on models and parameters in the Bayesian model averaging approach but also include a robustness check using Frequentist model averaging (Amini & Parmeter, 2012). One of the explanatory variables will be the time trend: given that the usage of coal is becoming pricier, both in terms of ecological impact and money, I hypothesize that the substitution of coal is becoming easier (as industries change their technologies and households change their preferences), which is reflected in the title of the thesis.


1. Introduction
• Motivation, contribution, and main findings
2. Introduction of topic
• Brief description of elasticity of substitution and interfuel substitution.
• Previous findings and estimates
3. Data collecting
• Search query and inclusion criteria
• Summary of studies
• Basic summary statistics of the estimates
4. Publication bias
• Argumentation for the presence of publication bias, methods of testing
• Visual, linear and non-linear models testing the presence of publication bias utilizing the relationship between the estimates and their standard error, other tests of publication bias using the distribution of p-values
5. Heterogeneity in the literature
• Choice of explanatory variables likely responsible for the heterogeneity across the estimates
• Model uncertainty, model averaging
• Best-practice estimate and discussion of the results
6. Economic implications
• Discussion of economic consequences
• Discussion of an impact of the estimates
• Recalculation of various consequences into monetary value
7. Conclusion
• Summary of the obtained results
• Implications for future research and drawbacks of this analysis
Charles University | Information system of Charles University |