![Sample our Social Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110822/TOC-Subject-Sample-2021-Social-Sciences.jpg"})
ABSTRACT
This study applies the PSM-DID approach to investigate whether the energy price has been impacted by the Russia-Ukraine war among OECD countries during 1999–2022. Compared with previous studies, we obtain robust and heterogeneous results by considering endogeneity from the perspective of quantitative policy assessment. It concludes the Russia-Ukraine war has significantly raised energy prices, leading to a 9% increase in energy prices in OECD countries. The heterogeneity results show that the Russia-Ukraine war impacted the energy prices of EU and NATO members but not those of non-EU and non-NATO countries. Additionally, only the energy prices in Southern and Western European countries have been significantly increased by the Russia-Ukraine war, in which Southern Europe has been the most affected, rising by 22%. These findings have great practical significance for analyzing the degree of influence of war on fluctuations in energy prices and geopolitical risks in regions.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
2 https://eu-solidarity-ukraine.ec.europa.eu/eu-sanctions-against-russia-following-invasion-ukraine_en
6 The use of DID model in this paper is the difference between the difference in energy prices between the treatment and control groups after the Russia-Ukraine war and the difference in energy prices between the treatment and control groups before the Russia-Ukraine war. An important prerequisite for the use of DID model is to satisfy the parallel trend hypothesis. It means the treatment group and the control group must have the same trend before the war, and the changes affected by other factors must be the same. Although the energy consumption prices of both the control and treatment groups are affected by the Russia-Ukraine war, the extent of the impact between the two groups are different. However, except for the Russia-Ukraine war, if the changes of the two sets of data are the same before and after the beginning time of the war, the results obtained imply estimates of the different effects of the Russia-Ukraine war on the two groups. That is, the result is that the Russia-Ukraine war affects the energy price of the treatment group relative to the control group, rather than the Russia-Ukraine war's impact on the energy price. Our subsequent empirical study verified the establishment of the parallel trend hypothesis. Therefore, it is feasible to analyze the treatment group and the control group by using the difference-difference method for the war evaluation in this paper. In this paper, the treatment group and the control group are distinguished according to the degree of energy dependence on Russia. Therefore, the empirical results suggest the change of energy prices in countries with high dependence on Russian energy imports compared with countries with low dependence on Russian energy imports due to the influence of the Russia-Ukraine war. Similar approaches could be referenced in Meng and Yu (Citation2023), which also divided countries into two groups based on energy dependence.
7 The energy consumer price index is the national consumer price index compared with the same period of the previous year in energy sector. It is available at the most aggregate level: CPI All Items (COICOP 01-12) with a breakdown in 12 COICOP Divisions and some additional aggregates. In this paper, the scope of energy consumer price index includes COICOP 04.5: Electricity, gas and other fuels plus COICOP 07.2.2: Fuels and lubricants for personal transport equipment. The types of energy include coal, solar, hydro, gas, fuels, lubricants and Other energy for heating and cooling according to Classification of individual consumption by purpose (COICOP) 2018.
8 Both GDP stock and GDP growth rate are important indicators reflecting economic growth. GDP growth rate focuses on the dynamic nature of economic growth, while GDP focuses on static evaluation. Compared with GDP stock, GDP growth rate can better reflect the actual situation of national economic growth speed, development potential and economic fluctuations, while GDP stock does not reflect the stage and time trend of economy. GDP growth rates are commonly found in many regular macroeconomic conclusions, such as Okun's Law. In addition, when a government sets an economic growth target, it often focuses on the speed of economic growth rather than the size of GDP in government documents. Also, the explained variable of this paper is the energy price index, which is also a dynamic indicator, and the GDP growth rate as the control variable is more compatible with it. Of course, it is perfectly possible to use the stock of GDP as a control variable.
9 It means the treatment group increased by 9% over the control group in OECD countries. Other results in the following text have similar implications.
10 The t value in can be calculated according to the regression coefficient and the corresponding standard deviation, and the t values of model (3), model (4) and model (5) are 3.42, 3.76 and 3.87, respectively.
11 Model (1) applies Mixed OLS and does not consider individual effects (i.e., fixed effects and random effects) in panel data, but regards panel data as cross-section data, and assumes that disturbance terms between different individuals are independent of each other. Model (2) just consider time fixed effect which captures impacts that do not change with individuals but change over time, such as macroeconomic cycles, etc. Model (3) just consider the individual fixed effect and finds the effects of differences between individuals that do not change over time, such as urban natural conditions, geographical features, and so on. Model (4) consider both time and individual fixed effect. The idea of Model (5) is to consider selecting the most representative period from each period before processing for single period matching on multi period panel data. The benchmark regression in Table 3 adopts a mixed matching method, which means that for each observation value in the treatment group during the treatment period (corresponding to each individual in the treatment group and each treatment period), the control group searches for the observation value with the closest propensity score for matching. Model (6) adopts the idea of phase by phase matching, which divides panel data into cross-section data phase by phase, and carries out phase by phase matching, and then merges into panel data.
Additional information
Notes on contributors
Tie-Ying Liu
Tie-Ying Liu, Associate Professor of the School of Economics and Management at Beijing Jiaotong University, China. He extensive professional experience has shaped his research interests in macroeconomics and energy economics. To date, he has made significant contributions to the field by publishing some papers in SSCI journals.
Chien-Chiang Lee
Chien-Chiang Lee, Professor of the School of Economics and Management at Nanchang University, China. His research interests include macroeconomics and financial economics. Professor Lee has contributed over 300 articles to academic journals. Professor Lee is also ranked 1% of the most influential authors in the field of economics and finance in Asia by IDEAS/REPEC ranking.