Council on Energy, Environment and Water Integrated | International | Independent

Understanding Segmentation in Rural Electricity Markets Evidence from India

Daniel Robert Thomas, Shalu Agrawal, S P Harish, Aseem Mahajan, Johannes Urpelainen
March 2020 | Power Markets

Suggested citation: Thomas, Robert, Daniel, Shalu Agrawal, S.P., Harish, Aseem Mahajan, and Johannes Urpelainen. 2020. Understanding Segmentation in Rural Electricity Markets: Evidence from India Energy Economics. Available at SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3531197


This paper develops a novel method for determining drivers of electricity demand without usage data. It employs a market segmentation design of Indian consumers from over 200 villages in four states. The consumers were clustered based on the willingness to pay, the usage level, and satisfaction with lighting. It then uses cluster membership as a dependent variable to determine which household-level factors predict electricity usage. This study applies machine-learning and more traditional regression techniques to determine the optimal number of segments, generate the segments, and determine the predictors of segment membership. The algorithm produces the number and characteristics of different segments based on the nature of the data itself in a systematic but inductive approach.

Key Findings

  • Indian electricity market can be segmented into three clusters based on households' willingness to pay, satisfaction with lighting, and appliance wattage.
  • The three clusters are (i) potential customers, (ii) low-demand customers, and (iii) high-use customers.
  • This differentiation allows identification of those who might be likely to demand more electricity in the future due to the potential increase in satisfaction and based on current usage.
  • Market segmentation can avoid the increased supply of energy to those who do not seek it, leading to a more efficient allocation of electricity overall.
  • Qualitatively, the three clusters represent different levels of potential and observed electricity usage.

Source: Milan George Jacob/CEEW

Potential customers

  • Individuals in this cluster currently use low levels of electricity and have little satisfaction with the quality of lighting. It indicates that an improvement in service to these customers could increase their demand.
  • Being a member of a scheduled caste or tribe increases the likelihood of being in this cluster.

Low-demand customers

  • In this cluster, current usage is low, but satisfaction with lighting is high. These customers are likely to remain at their current usage levels regardless of improvements in service.
  • The likelihood of being in this cluster increases with an increase in the age, preferences for cheap goods, and having a grid or mini-grid connection.

High-use customers

  • The third cluster consists of customers with both high satisfaction and high usage; this cluster is likely to exhibit low growth in demand in the future.
  • Attending school and having acquired higher education, having a large household, having leadership traits, and being risk-averse increase the likelihood of being in this cluster.
  • Meanwhile, being in a scheduled caste or tribe lowers the possibility.
  • An increase in household expenses, hours of village grid electricity, and household grid connections increase the likelihood.
  • In places where a large number of customers exhibit characteristics similar to those in the Potential Customers segment, a focus on increasing the quality of electricity service can significantly increase demand.
Attending school and having acquired higher education, having a large household, having leadership traits, and being risk-averse increase the likelihood of being in the high-use consumer cluster.

Sign up for the latest on our pioneering research

Explore Related Publications