Real estate players are preparing for a real upheaval. The reason ? The entry into force of a series of reforms intended to stimulate the energy renovation of buildings. Since 1er last July and the total overhaul of the DPE, all owners must, for example, append the estimate of the energy consumption and the carbon emission rates of the accommodation to the sales contract or rental lease (more than four months). From September, an energy audit must be provided for housing classified F or G, in addition to the DPE. But it is undoubtedly the “Climate and Resilience” law of August 22, 2021 that will generate the strongest shocks on the market. Considered “indecent”, some 600,000 housing units (classified G) will be prohibited from rental from 1er January 2025. This will then be the case in 2028 for F-class housing (1.2 million housing units) and in 2034 for E-class housing (2.6 million). Next August, landlords of F and G classified accommodation will be prohibited from raising their rents.
An unprecedented problem
This new legislative framework will inevitably impact the properties – and the borrowers – presenting a bad DPE. Let us imagine several scenarios. A borrower wishing to take out a bridging loan in 2025, but does not have the financing to renovate his home, and therefore leave it… Another who, following the loss of his job, finds himself unable to rent his home… If a significant proportion of the loans granted by a bank relate to misclassified properties, the risk of a discount on the value of the properties in its portfolio is high.
Consequently, these new obligations affect the very heart of the banking business: the ability of institutions to manage and control their risk exposure. Only a precise vision of their portfolio will allow them to operate the appropriate provisioning. However, in reality, while banking establishments have information related to DPEs on their recent credit flows, the thing is much less true for loans granted several years earlier.
Machine learning, to estimate its DPE
Logically, a number of banking groups want to be proactive in analyzing the real estate in their portfolio. How can you have good visibility on the energy balance sheet of the housing units that make up a loan portfolio? To this strategic and complex question, only artificial intelligence is capable of providing an answer. The promise: to give, regardless of the input given and the level of information available, an estimate of the energy classification of each home, associated with a level of confidence.
Concretely, to propose an estimate of the DPE, the algorithm of machine learning starts from very poor quality information (postal address, surface area, type of accommodation), before qualifying it and cross-checking it with various data. The main strengths of the technology lie in its ability to cross-check all available sources of information: years of construction, public DPE databases, analysis of the specific characteristics of a dwelling, comparison of properties with similar characteristics to the within a geographical area and ability to extrapolate using image analysis.
Make no mistake, this is the estimate of the DPE of a dwelling accompanied by a confidence index. Still, the creation of an algorithm from pre-existing bases, capable of being enriched by internal and external data, presents a very strong added value. All in a very limited time: once the algorithm has been created, a few weeks are enough for the AI to be able to process several tens of thousands of lines of credit. An unparalleled lever available to credit institutions to get as close as possible to the reality of their financial portfolio. And thus measure their exposure to risks.
We wish to thank the writer of this short article for this remarkable content
DPE: AI, to better identify risks
Explore our social media profiles as well as other pages related to themhttps://www.ai-magazine.com/related-pages/