Electricity Infrastructure

In this section, we discuss the evidence on various factors affecting the demand for electricity services – such as electricity prices, other drivers of overall appliance ownership and use (e.g. household wealth, income, and temperature), and the adoption and use of energy efficient technologies specifically.

Building on the earlier discussion of reliability and electricity quality, it is worth noting that some demand-side interventions might affect the functionality of the infrastructure itself. For example, increases in demand at peak times (i.e. extreme temperatures inducing greater use of cooling or heating services) may lead to outages in the distribution system, either due to loadshedding or overloads resulting in breakage.

In addition to the challenges of providing adequate quality electricity to existing customers, utilities must also invest in infrastructure to meet future demand. Predicting future demand is challenging, particularly in LMICs, as it involves predicting changes on both the extensive margin (i.e. increased demand from newly connected customers) as well as on the intensive margin (i.e. changes in demand among existing customers). Demand among existing customers may change as a result of adopting new appliances, replacing old appliances with more efficient versions, or by using their existing appliances more.

Electricity pricing

Electricity pricing and tariff reform are tools available to affect consumption in multiple ways. As discussed in Section 3, tariff changes that are intended to close the gap between the cost of supplying each unit of electricity and the price paid per kWh delivered can affect consumer behaviour. In this section, we aim to cover the evidence on changes to electricity prices from a different perspective than those discussed in Section 3. For example, in settings with newly electrified households, prices may be set in an effort to stimulate electricity demand through the ownership and use of new and additional appliances. This is particularly true in locations where utilities have invested in increased generation capacity or transmission and distribution infrastructure with the expectation that demand will increase in the future.

In other settings, policymakers may reform electricity tariffs because they seek to induce the deployment of distributed renewable generation and/or the electrification of certain services, such as cooking or transportation (via electric stoves and vehicles). McRae and Wolak (2021) study such a policy, by modeling the potential effects of a new pricing structure in Colombia that involves setting fixed charges based on the customers willingness to pay at the marginal cost.

In other cases, utilities might need to limit demand, often during hours of peak consumption. It is easy to imagine a scenario in which the utility wants to see demand increase overall, but must also ensure it does not increase too much during peak hours to avoid overloading the transmission and distribution networks. Different kinds of interventions could aim to affect demand in these different ways. Khanna et al. (2024) provide some of the first experimental evidence testing financial incentives to shift demand over time in an LMIC.

Drivers of electronic appliance ownership and use

Within this subsection we present the evidence on some of the commonly discussed drivers of appliance ownership and use – other than electricity prices – that thereby affect electricity demand, including household income and temperature, as well as the electrification of end uses (e.g. cooking and transportation).

Income, wealth and increasing temperatures

Some recent literature seeks to understand electricity demand among relatively lower-income households, as these are the households that are typically the last to be electrified (Lee et al. 2020b, Masselus et al. 2024). In both cases, the authors find low uptake of appliances and low demand for electricity, indicating that connections to the electric grid are not the only constraint limiting households. These findings indicate that we cannot expect immediate demand upon connecting new customers to the grid. When considering the financial scenarios for future grid expansion, electricity utilities should expect that new connections may bring low demand customers with low ability to pay (Burgess et al. 2020, Bajo-Buenestado 2021). With that in mind, policymakers may consider promoting off-grid renewable technologies (covered in Section 5) instead of grid extensions – particularly when electrifying those locations that are remote and therefore the most costly to connect. These off-grid technologies may be more cost-effective options for lower demand consumers.

There is evidence that households invest in more electric appliances as they grow wealthier; however, this relationship is nonlinear. Demand for appliances can be particularly sensitive to credit constraints, particularly in the case of large appliances (Wolfram et al. 2012, Gertler et al. 2016). Further, we are starting to understand the important interactions between household income and temperature changes in the adoption of air conditioners, which are an important determinant of electricity demand. Through an empirical analysis of 25 countries that account for approximately two thirds of the world’s population, De Chian et al. (2025) find that households owning ACs consume 36% more electricity than those without ACs, on average; however, this also varies by weather and income (De Chian et al. 2025).

Currently, there are billions of people living without air conditioners in low- and middle-income countries (LMICs) with tropical climates (Biardeau et al. 2020). Around the world, air conditioner ownership has increased with both gains in household income and temperatures (Auffhammer 2014, Davis and Gertler 2015, Davis et al. 2020, Randazzo et al. 2023). This appliance is particularly important as it has been shown to be crucial in mitigating the effects of extreme heat in high income countries (see for example, Barreca et al. 2016). The concern, however, is that adoption of air conditioners in LMICs is expected to be only among the wealthiest households in those countries (Davis et al. 2021).

This begs the question: how will the poorest households adjust to extreme heat? If these households have a formal grid connection they typically consume low quantities of electricity and - if an increasing block price exists - are on the bottom tier of the price schedule (in the so-called “lifeline” tariff block). Poor households therefore not only face the high upfront costs of purchasing an AC. The recurring costs to run an AC are also a constraint. If the household is using the AC regularly and through a formal grid connection, this would push the household out of the lifeline tariff block to a higher pricing tier.

Ahmad et al. (2025) investigate household response to extreme heat in Pakistan and find evidence that, while both billed and unbilled consumption (electricity theft) increase with hotter temperatures, the unbilled consumption is more responsive to temperature and has a sharper increase with hotter days. This indicates households are likely adjusting to hot temperatures by stealing more electricity. Among those households formally connected to the grid, they find evidence that wealthier households maintain a steady consumption of electricity from the grid, whereas poorer households appear to cut back consumption on moderate days.

Decarbonising cooking

As part of efforts to decarbonise certain sectors, there are attempts to shift households to electricity for cooking and transportation. Increasing the use of electric cookstoves is a particular focus in energy and health circles as a way to limit exposure to household air pollution, and potential reduce greenhouse gas emissions while also increasing electricity demand (Clasen et al. 2024, Pattanayak et al. 2019, Gould et al. 2023). Although Gould et al. (2023) find an increase in electricity consumption among those who shift to cooking with electricity, there are a number of barriers to making this transition. As with other appliances, household income and credit constraints play an important role in determining adoption of electric stoves.

It is also important to acknowledge how these increases in demand might affect the electricity system more broadly if adoption of electric stoves and vehicles occurred on a large scale. Power quality and reliability issues may result from high consumption, especially during peak periods (Jacome et al. 2019, Carranza and Meeks 2021).

Demand for electricity and energy-efficient technologies

Fowlie and Meeks (2021) review empirical evidence on the economics of energy efficiency in LMICs, including market failures and barriers to the adoption and use of energy-efficient technologies. In the following sub-sections, we note aspects of particular relevance to policymakers working in LMICs.

Adoption of energy-efficient technologies

Energy-efficient technologies offer a range of potential benefits that are interconnected with electricity infrastructure. Theoretically, they can reduce the strain on an electricity grid at times of peak demand, which could improve reliability for all customers, especially in the face of rapidly growing electricity demand (Ahmad and Zhang 2020). They can also reduce overall consumption, which has important climate implications in places where electricity generation is carbon-intensive. However, research on the interaction between energy efficiency and electricity infrastructure remains limited. While some studies (e.g. Carranza and Meeks 2021, Meeks et al. 2023) directly study these linkages, most of the literature focuses on energy efficiency in a more isolated manner.

Much of the electricity-specific causal evidence on energy efficiency in LMICs focuses on energy-efficient lightbulbs. Toledo (2016) conducted an experiment in Brazil to assess the impacts of three subsidy levels and environmental messaging on the uptake of LEDs. The households offered free LEDs adopted them almost universally, and those with the smallest subsidy do not have significantly higher adoption when also receiving environmental messaging, but those in the middle-level subsidy group increase the uptake of LEDs by 20% when they receive environmental messaging. Beattie et al. (2022) experimentally elicit household willingness to pay for LEDs, which they find to be higher after being provided information on the energy savings from LEDs. There are also descriptive studies aiming to understand how knowledge (Asif et al. 2021) and gender (Aziz et al. 2024) might drive the uptake of energy-efficient appliances more broadly.

Use of energy-efficient technologies

A smaller set of studies focus on the impacts of energy-efficient appliances on electricity consumption. Energy-efficiency improvements can potentially expand access to energy services, increase productivity, and reduce the strain on the grid depending on their use and the device that they are replacing (if any); however, they also can result in higher overall consumption due to rebound effects or infrastructure-driven changes in consumption behaviour. Although the number of studies is limited, those that do exist highlight the complex interactions between efficiency and electricity infrastructure, with results often diverging from the anticipated reductions in consumption.

In Mexico, Davis et al. (2014) found that refrigerators that were replaced with a more efficient model only reduced consumption by a quarter of the expected amount based on engineering estimates and that consumption actually increased when air conditioners were upgraded to a more efficient model. This is a classic example of a rebound effect whereby households are using the more efficient appliances more due to lower operating costs.

Within a firm setting, Ryan (2018) found that energy consulting provided to Indian manufacturing firms led to an increase in electricity consumption. Specifically, he highlights that by integrating energy efficiency improvements at key points to complement skill and capital in the production process, firms were able to become more productive but ended up increasing electricity consumption overall.

Through a randomised experiment in the Kyrgyz Republic, Carranza and Meeks (2021) study the distribution of energy efficient lightbulbs to households served by different transformers; through the experimental design some transformer areas had a higher proportion of households receiving the energy efficient bulbs than others. They found evidence that in high-saturation areas, electricity reliability improved as the bulbs shaved load sufficiently to alleviate strain on the transformer during peak hours, allowing electricity consumption to increase.

Evidence gaps and future directions of research

There is increased attention in the literature on both the drivers of initial adoption and increased use of energy efficient appliances, but significant gaps remain in understanding how the increased use of these appliances interact with electricity infrastructure over time and as demand grows. There is limited research on how energy efficiency affects the quality and reliability of power systems by reducing load especially during peak hours, and conversely, how quality and reliability might influence the uptake of these appliances in the first place. These interactions are likely to unfold in dynamic ways, potentially over longer time scales, suggesting the need for more longer-term research that aims to capture these relationships more clearly and to provide deeper insight into the long-term effects of, in particular, large-scale investments in energy efficiency.

Additionally, much of the existing literature on energy efficiency focuses on urban and/or firm settings. Given that transmission infrastructure in rural areas is usually unable to handle large loads, these areas may be particularly susceptible to small changes in demand that could come from increased use of energy efficiency appliances.

Reframing impacts within the reliability-consumption cycle

As discussed throughout this section, the reliability-consumption loop is central to how electricity interventions play out. Efforts to improve reliability or reduce utility losses could have impacts that reverberate across the broader electricity system.

Some more recent research provides useful examples of the ways in which studying multiple types of electricity infrastructure interventions – both demand- and supply-side - can more clearly account for these interconnected processes.

The dynamics found in the existing research, such as Ryan (2018) and Carranza and Meeks (2021), underscore the need for more research that studies the interplay between energy efficiency, electricity infrastructure and productivity, rather than treating them as isolated topics. The papers highlight how demand-side interventions like the promotion of energy efficient appliances can have unintended consequences on consumption, which could in turn create new pressures on the grid. A complementary example, in the context of Pakistan, is Ahmad et al. (2024). As previously discussed, the paper provides evidence on the impacts of installing theft-resistant cables. The authors find lower rates of theft following the installation of the cables, but also improvements in reliability. This in turn shows how a supply-side intervention can also affect the inner reliability-consumption loop, albeit through a different mechanism that the one described in Carranza and Meeks (2021).

In order to continue to unpack these complex relationships, future research should move away from studying just demand or supply side factors in isolation without considering broader impacts.

For full reference list see the end of the conclusion chapter.