The availability of data on the energy efficiency of properties in Italy is crucial both for evaluating public policies aimed at achieving the European objectives to decarbonise the real estate sector and for use by banks required to estimate the impact of transition risk on the guarantees in portfolio and to comply with disclosure obligations required by regulation. This work provides a contribution to the discussion on the estimation of energy efficiency data.
Using an algorithm known as Random Forest Classifier, based on a limited number of variables, it is possible to estimate the energy classes of Italian properties. The procedure suggests that the share of energy-inefficient houses (those in energy class F or G) is much higher than that what can be inferred from a sample of the Sistema informativo sugli attestati di prestazione energetica (SIAPE). In particular, considering a margin of error of one class, it is estimated that energy-inefficient homes are around 80 percent of existing stock.
