Cost of CO₂ emission mitigation and its decomposition: evidence from coal-fired thermal power sector in India

India is able to electrify all the villages in April 2018, yet about 20% household have no access to electricity (http://​saubhagya.​gov.​in/​ as accessed on May 15, 2018). A village is said to be electrified if electricity is provided to public places and at least 10% of the total number of households are electrified in the village (http://​www.​ddugjy.​gov.​in/​portal/​definition_​electrified_​village.​jsp, accessed on 17.02.2019).

Under the Kyoto Protocol, India was not required to reduce carbon emissions, but at the Paris Agreement, India has pledged to reduce the CO₂ intensity of GDP by about 30–35% by 2030 relative to 2005. Reduction in the desired level of the intensity requires India to take some regulatory measures.

Carbon or CO₂ mitigation cost (CMC) and pollution abatement cost (PAC) words are used interchangeably throughout the paper.

Martin et al. (1990) and Bellas (1998) consider pollution abatement activities to be independent from marketed output production processes.

\({\text{CMC}} = \left( {{\text{CMC index}} - 1} \right) \times {\text{Electricity Output}}\).

Statistical Yearbook 2018, Ministry of Statistics and Programme Implementation, accessed from mospi.gov.in.

Information on thermal power plants is available on financial year basis in India, starting April of a year and closing in the March of following year. Therefore, 2005 refers to April 2005–March 2006 and 2013 refers to April 2013–March 2014.

The Clean Energy Cess has been replaced by a GST Compensation Cess at the rate of INR 400 per metric ton of coal and lignite with effect from July 01, 2017.

Considering the prevailing technology as unregulated technology makes the estimates of CMC upward biased (Färe et al. 2003).

Descriptive statistics in Table 2, reveals that electricity production and CO₂ emissions were about 12.14 and 9.03% respectively, higher in 2012 compared to 2008, implying declining carbon intensity of electricity generation. If the intensity was constant over the period then an average plant had produced 216 thousand tons more of CO₂ than actually produced, i.e. 9726 thousand tons of CO₂ emissions were mitigated by sacrificing 3.5% of electricity production by the sample 45 plants.

Zhou et al. (2008) review the literature on the use of DEA to model environmental performance.

A given technology captures the basic relations between inputs and outputs, based on physical and natural laws. From the technology point of view, emission is not a freely disposable output. It is costly in terms of the good output foregone or in terms of more inputs required to produce same level of good output, i.e. there is a positive trade-off between good output production and emission generation. The prevailing technology is of weak disposability. Strong disposability is a counterfactual case, considered to compute the cost of emission reduction. There are two approaches for modelling free disposability of bad outputs. We follow Färe and Grosskopf (1983) and other subsequent studies in modelling the free disposability of bad outputs, by treating bad outputs equivalent to good outputs. The second approach drops the constraint related to the bad outputs in the maximization of good output or measure of technical efficiency (Färe et al. 2016). The difference between these two approaches of modelling free disposability of bad outputs is inclusion of downward sloping frontier of bad output. Färe et al. (2016) follow the later approach, which does not involve bad outputs under free disposability of bad outputs formulation.

Right to Information (RTI) Act 2005 mandates time bound reply to citizen appeals for government information (http://​righttoinformati​on.​gov.​in/​).

An environmental production function is a special case of an environmental directional distance function, which credits for the expansion of good output. This formulation has been chosen as it replicates Indian CO₂ mitigation policy in thermal power sector.

In the measurement of technical efficiency of thermal power plants in India, Singh (1991), Shanmugam and Kulshreshtha (2005), Shrivastava et al. (2012) Sahoo et al. (2017) also assume constant returns to scale (CRS). Some studies measure technical efficiency under variable returns to scale (VRS) by adding a convexity constraint to the CRS model. However, adding a convexity constraint to the CRS model under weak disposability is not equivalent to a VRS model under weak disposability (Färe and Grosskopf 2003). Chen (2013) indicates that VRS model under weak disposability condition is highly nonlinear and is difficult to solve, and the production set is non-convex and non-monotonic.

Technologically there is a positive relation between the production of good and bad outputs, irrespective of the state of regulation. Under regulation, to internalize the emissions effect, the good output is reduced for reducing emissions and in an unregulated situation more of good output is produced simultaneously producing more of bad outputs. However, we use free disposability condition as a counterintuitive case (Färe et al. 2016).

Net electricity generation is defined as gross electricity generation minus auxiliary consumption of electricity which is used by the plant for generation of electricity.

CO₂ Baseline Database for the Indian Power Sector, User Guide, Version 11.0, April 2016, CEA.

Domestic coal, used in thermal power plants in India, is assumed to be homogeneous, almost having same heat content. Since plant-wise data on coal quality and heat rate was not available for the study period, consumption of coal has been used. In Indian thermal power plants, use of non-coal fuel is minimal. The plants use oil, only as an ancillary fuel.

Dhrymes and Kurz (1964), compute capital input as a product of total capacity available during the year, its operational availability factor and number of hours of in a year, measured in Gigawatt hours. For details on the data and variable measurement, please see Jain and Kumar (2018).

Wang et al. (2018) find that CMC is the product of carbon productivity and output elasticity of substitution of CO₂ emissions. Output elasticity of substitution (OES) is defined as a ratio of changing rate of the frontier’s desirable output level due to the changing rate of CO₂ emissions, i.e. \({\text{OES}} = \frac{\% \Delta y}{\% \Delta b}\). OES indicates the substitution relationship between the desirable output and CO₂ emissions. For a plant having higher carbon productivity, the CMC would be higher since the possibilities of substitution will be lower in comparison to a plant having lower carbon productivity.

Year-wise plant level sale price of electricity at current prices was taken from Central Electricity Authority (CEA) and was converted into 2004–2005 level prices, using fuel price index of Reserve Bank of India. The losses are converted to $US, to make it understandable to readers globally.

The terms sub-critical, supercritical, and ultra-supercritical are related to steam operating conditions in the boiler of a plant defined in terms of pressure and temperature. The main steam pressure (MPa) is less than 22.1 for the sub-critical plants, it lies between 22.1 and 25 for supercritical plants and for ultra-supercritical plants this value is higher than 25. Ultra-supercritical and supercritical technologies are more efficient, require less fuel per unit of electricity generated and produces less emissions relative to subcritical plants. In India, generally the power plants in the capacity of 100–600 MW capacity are sub-critical and of greater than 660 MW capacity are supercritical.