Page 17 - ECLECTIC MARCH-2022 INSIDE PAGES_p2.indd
P. 17
rease the level of imports of coal, imported coal is the explanatory (independent) variables and response
used majorly for blending and comprises a small portion (dependent) variables. The days of stock of coal with
of days of coal stock. Similarly, rising overdues to gencos power plants in India is the variable of interest and forms
and subsequently to coal suppliers may not have led our dependent variable. The despatch volumes of coal,
to curtailment of coal supplies as long as the gencos the volume of imported coal, overdues to gencos and the
managed to meet their working capital requirements demand of power forms our independent variables. The
through debt or other means. There is also a slew of variables are normalized prior to the regression analysis,
nationalised players of disproportionate size in both so that the mean of each of the resulting normalized
power generation and coal mining and marketing areas. datasets of the corresponding variables is 0 and the
In the subsequent section, we attempt to study the standard deviation is 1. Thus, the regression equation we
relationship between the overall level of coal stock in attempt to solve for is as follows:
thermal power plants across India and the indicators for
each of the reasons listed in the section above. Stock = β0 + β1 Demand + β2 Despatch + β3 Import +
β4 OD + e
To study the effect of the various reasons listed in Stock = Normalized Days of Coal Stock at Power
the previous section, the data of days of coal stock is Plants across India at the end of the month
collected from National Power Portal (NPP). The coal
stock available across power plants in India at the end Demand = Normalized Power Demand for the
of the respective months is taken as the coal stock for month in Million Units
the month. The power demand data (in Units, kWh) is
extracted from the website of Central Electricity Authority Despatch = Normalized Monthly Volume of Coal
(CEA). Data of despatch of coal by Coal India Limited to despatched by Coal India Limited and other captive
the power sector, and volume of coal imports has also mines to power sector
been taken from the website of Ministry of Coal. The data
has been collected for the months of the pandemic and Import = Normalized Monthly Volume of Imported
disruption of economic activity, starting from April 2020 Coal by power sector
to October 2021. As a proxy of overdues to Coal India
from the power sector, data for overdues from discoms OD = Normalized Total outstanding dues to
to gencos has been considered. This data for total generating companies from distribution companies
overdues to gencos has been collected from the PRAAPTI at the end of the month
(Payment Ratification And Analysis in Ministry of Power,
Government of India, Power procurement for bringing β0 = Constant term in the model
Transparency in Invoicing of generators) database. βi’s = Slope coefficients of each explanatory variable,
To estimate the dependency of the coal stock with each
of the other factors, we use linear regression with multiple e = Error term of the model, also called the residuals
variables. Multiple linear regression (MLR), also known
simply as multiple regression, is a statistical technique A multiple linear regression works on the principle
that uses several explanatory variables to predict the that for the best fit model, the sum of squared errors is
outcome of a response variable. The goal of multiple linear minimum. Statistical MLR software and add-ins helps find
regression is to model the linear relationship between all of the ’s in the multiple regression. In these models, the
null hypothesis is that the βi’s = 0, which is to say that the
corresponding independent variable does not influence
the dependent variable. If the p-value of corresponding
variable in the regression output is less than the level
of significance, we can reject the null hypothesis and
conclude that the independent variable is significant in

   12   13   14   15   16   17   18   19   20   21   22