*H4b: GDP and DE are positively causal related*

Some researchers have reached an opposing conclusion that energy subsidy reform would produce positive results. Steenblik and Coroyannakis [30] used the computable general equilibrium (CGE) model to simulate the positive effects of removing coal subsidies in West‐ ern European countries, such as promoting the industrialization of the power sector and in‐ creasing coal production and exports. United Nations [3] concluded that cutting energy subsidies could have significant impacts on residents, although this requires a more indepth analysis in the future. Conversely, some researchers believed that fossil energy reform would increase energy use efficiency and household income levels. Choi, Roh and Yoon [9] indicated that increase in energy price could improve energy efficiency significantly. Thus, the energy price mechanism is at the core of energy reform, and energy subsidies are crucial determinant of energy prices.

Anderson and Leach [27] showed that energy subsidies in the United States would impede the use of new energy and reduce energy use efficiency. Shah and Larsen [31] showed that if the total energy subsidies worth almost \$230 billion in 1990 could be removed, CO2 emis‐ sions worldwide would decrease by 9.5%. Using the global coal model, Lam and Shiu [32] analyzed coal subsidy reform in Japan; the results showed that removing the coal subsidies in the power supply and industrial boiler sector would reduce global CO2 emissions by 0.2%. The IEA [5] also indicated that global CO2 emissions would decrease by more than 6% by 2010 if the fossil energy subsidies in the power sector were removed. We use these re‐ search data to test and verify these countries, and the relationship between subsidies policy and efficiency. Therefore, Hypotheses 5a and 5b are as follow:

*H5a: Verify that causal relationship between subsidy and OE*

*H5b: Verify that causal relationship between subsidy and DE*
