Environmental Tax Policy Review of Romania

Annex D. Simulation model

To estimate the fiscal implications of the tax reform, the analysis adopts an isoelastic model, following the methodology of Stretton (2020[1]). Under this specification, the energy use in the reform scenario is a function of the relative price change, governed by a constant price elasticity of demand1:

$Q_{i,Reform}= Q_{i,WEM}*{\left(\frac{P_{i,Reform}}{P_{i,WEM}}\right)}^{{\epsilon }_i}$

• where $P_{i,WEM}$ and $P_{i,Reform}$are the total price of energy source I under the baseline and the reform scenarios respectively;

• $Q_{i,WEM}$ and $Q_{i,Reform}$ are the corresponding quantity of energy i;

• and ${\epsilon }_i$ is the price elasticity of demand for energy source i.

Price elasticities capture the behavioural response of consumer with respect to price changes. Short-term elasticities predominantly reflect adjustments in fuel demand conditional on existing fuel technologies and the prevailing energy efficiency of housing stock. In contrast, medium- and long-term elasticities incorporate households' broader behavioural and technological responses, such as energy source switching (captured through cross-price elasticities) and investments in energy efficiency upgrades, including housing retrofits. Due to the lack of price elasticities for Romania, the main analysis is based on the short and long-term elasticities as estimated in the meta-analysis by Labandeira et al. (2017[2]).

For simplicity, the model does not include cross-price elasticities. Therefore, it does not capture fuel substitution effects, such as households switching from coal to natural gas, in response to relative price changes. A more comprehensive specification would account for such behavioural adjustments. Nonetheless, given prevailing uncertainties in baseline energy demand, the model offers a approximations of first-order impacts.

For each scenario, the consumer price of energy source i is computed as:

$P_i=\left({Price before taxes}_i+{Excise Tax}_i+{CO2 Tax}_i+{Pollution Tax}_i\right)*(1+{VAT}_i)$

capturing the price change from an excise tax, a carbon tax, an air pollution tax, and the VAT.

To simulate the environmental impacts of the reform as well as to include a carbon and pollution component within the taxes, the model retrieves both CO₂ emissions and air pollution through emission factors associated with combustion. Fuel use data, converted into kilowatt-hours (kWh) using standard calorific values, enables the estimation of emissions using the following formulas:

$E_i= Q_i*{EF}_i$

$A_i= Q_i*{AF}_i$

where:

• $E_i$ and $A_i$ represent the emissions of CO₂ and air pollutants (PM_2.5) of energy source i, respectively;

• ${EF}_i$ and ${AF}_i$ denote the CO₂ and air pollution emission factors2 per unit of energy.

The fiscal impact of the reform scenarios is quantified through the calculation of tax revenues generated from multiple sources:

${Revenue}_{Reform}=\sum _i{excise}_{Reform}*Q_{i,Reform}+\sum _i{carbon tax}_{Reform}*E_{i,Reform}+\sum _i{vat}_{Reform}*Q_{i,Reform}*P_{i,pre-VAT,Reform}$

In countries that implement a VAT, the tax is typically levied on a product’s final market price (Stretton, 2020[1]). If a country imposes a carbon tax, the final price includes the cost of the carbon tax for covered products. To avoid creating implicit subsidies and to preserve tax neutrality, it is essential that VAT be applied uniformly at the standard rate to the post-carbon-tax price of all VAT-eligible goods and services.

The net fiscal impact is derived as the difference in total revenue between the reform and the baseline scenario:

$∆ Revenue= {Revenue}_{Reform}-{Revenue}_{WEM}$

As robustness check, the simulation includes cross‑price effects and replaces the single own‑price elasticity with an elasticity matrix as below using a constant-elasticity demand specification:

$Q_{i,Reform}= Q_{i,WEM}*\prod _j{\left(\frac{P_{j,Reform}}{P_{j,WEM}}\right)}^{{\epsilon }_{ij}}$

The own‑price is captured when i=j. Given the absence of estimations within the Romanian context, the model assumes a cross-price elasticity of 0.02 across energy products.

References

[3] EEA (2023), “EMEP/EEA air pollutant emission inventory guidebook 2023”, https://www.eea.europa.eu/en/analysis/publications/emep-eea-guidebook-2023.

[2] Labandeira, X., J. Labeaga and X. López-Otero (2017), “Energy demand elasticities for OECD countries: A meta-analysis”, Energy Economics, Vol. 65, pp. 233–238.

[1] Stretton, S. (2020), “A simple methodology for calculating the impact of a carbon tax”, World Bank.

Notes

← 1.  The functional form used is an isoelastic demand function $Q\left(P\right) = Q₀ *{\left(\frac{P}{P_0}\right)}^{{\epsilon }_i}$. Differentiating with respect to P gives: $\frac{dQ}{dP}=Q_0*\frac{{\epsilon }_i}{P_0} *{\left(\frac{P}{P_0}\right)}^{{\epsilon }_i-1}$. Substituting into the definition of the price elasticity of demand:${\epsilon }_i=\frac{dQ}{dP}*\frac{P}{Q}$ yields ${\epsilon }_i=\frac{Q_0*{\epsilon }_i}{P_0} *{\left(\frac{P}{P_0}\right)}^{{\epsilon }_i-1}*\frac{P}{Q_0*{\left(\frac{P}{P_0}\right)}^{{\epsilon }_i}}$ , and simplifying ${\epsilon }_i={\epsilon }_i$. This confirms that the elasticity remains constant and equal to εᵢ across the price range.

← 2.  Air pollution emission factors are retrieved from EEA (2023[3]).