As ever more countries commit to net-zero greenhouse gas emissions in official policy documents and laws, a series of relevant building energy policies have emerged in different areas to enhance building energy usage efficiency. However, the quantification of their effectiveness is challenging due to substantial uncertainty. Much of relevant research employ linear regression models to analyse the correlation between time and energy consumption, aiming to ascertain the energy savings linked to a specific policy. Nonetheless, these approaches may deviate to some extent due to priori unknown or underestimated interventions. In this study, we develop econometric models to investigate the causal link between energy savings and building benchmarking policy in Singapore for office buildings. Using matching and regression techniques, we find that Singapore’s 2019 benchmarking policy contribute to 14% energy savings across all office buildings in 2020. However, we also find that the policy does not reduce energy consumption homogeneously in different buildings, exhibiting a more pronounced impact on buildings that are not certified as green buildings prior to 2019. These buildings have the potential to reduce their energy use intensity (EUI) by 19% if they can get certified in 2019, which is equivalent a reduction of 36.4 million dollars per year in electricity costs based on average electricity prices. These findings enable a better understanding of benchmarking policy impact and provide valuable insights for policymakers to conduct investment analyses related to building energy.