Association for Public Policy Analysis & Management

Motivation: Global electricity generation is a major contributor to greenhouse gas emissions, as most economies rely heavily on fossil fuels for electricity production (Myhrvold and Caldera, 2012). In South Africa, coal-fired power stations account for over 90% of the electricity generated (Winkler 2011). Together with its electricity being among of the world’s cheapest (City of Cape Town, 2010), this contributes to South Africa currently being ranked among the top 12 emitters of carbon dioxide in the world (Global Energy Statistical Yearbook, 2017).

In 2010, South Africa committed itself to reducing greenhouse gas emissions by 34% by the year 2020 (Winkler et al., 2010), making energy conservation a top priority for both national and local governments. In the Western Cape Province – the country’s fourth-largest by population and second largest by economic size – energy conservation efforts must address electricity usage in the City of Cape Town, which accounts for over half of provincial energy consumption and greenhouse gas, with further challenges resulting from rapid urbanization and energy insecurity resulting in periodic “loadshedding” or brownouts.

Intervention: Using extensive qualitative interviews, we worked with the government of the Western Cape to design nudges leveraging social comparison and assignment of responsibility aimed at reducing electricity consumption in a large provincial government office building with 24 floors. These nudges were tested in a randomized-control trial over the period June 2015 - October 2016. Floors in the first treatment arm received general energy conservation information emails and participated in weekly inter-floor competitions, while floors in the second treatment arm received both energy conservation information, participated in weekly inter-floor competitions and were also assigned a weekly energy “floor advocate”.

Methodology: The experiment was conducted in 4 Dorp Street, a large office building in Cape Town with twenty-four floors, where several provincial government departments are headquartered. Two smart meters each, were installed on all floors in the building prior to the start of the study. These meters were closely monitored for two years while correcting for problems such anomalies in meter readings, meter breakdowns and tracking of floor inventories. At the end of the testing period, only twenty-one floors of the twenty-four floors. Floors were assigned to two treatments arms and a control group, with seven floors in each group.

Results: Our interventions led to large declines in electricity consumption, with a 9% reduction (p < 0.10) due to Treatment 1, and a 14% reduction (p < 0.05) in energy use from Treatment 2 . In addition, larger treatment effects are observed in non-work compared to work hours, which is consistent with conservations efforts driving the results.

Conclusion: These reductions – which are among the largest demonstrated in any setting – cause us to re-evaluate the conventional wisdom that asserts that it is harder to nudge behavior in non-residential settings (such as office buildings) where users do not face the financial consequences of their behavior than it is in residential settings, where they benefit financially from any conservation efforts.