Do Mandatory Energy Audits Reduce Building Energy Use? : A Bayesian Analysis of New York City's Local Law 87

Cities are experimenting with a range of policy and regulatory tools to reduce energy use in the building sector as a pathway to substantial carbon emission reductions. Increasingly, cities are turning to policies that reduce information asymmetries and knowledge gaps through data transparency, focusing on existing buildings and the potential to achieve up to 50% energy savings through retrofit adoption. These include energy disclosure - or benchmarking - ordinances and, more recently, mandatory energy audit requirements. While energy benchmarking is a relatively low-cost mandate, energy audits impose non-trivial costs in terms of capital and time for building owners. The question we explore here is whether mandatory energy audits lead to reduced energy consumption over time. We collect detailed energy use and building attribute data from 2011 to 2016 for approximately 4,000 buildings in New York City and integrate with a unique dataset of mandatory energy audit reports for a randomly selected sample conducted in 2013 and 2014. We use non-parametric statistical tests and Bayesian regression to evaluate the impact of energy audits on building energy performance, compared to a control group of matched buildings not subject to the audit requirement. We find that mandatory energy audits have a modest negative impact on energy consumption, consistent with savings expected from low-cost energy conservation measures and retro-commissioning activities. The result reinforces the hypothesis that audits, by themselves, create an insufficient incentive to invest in energy efficiency upgrades at the scale needed to meet city carbon reduction goals