*Commercial Real Estate (CRE) is a broad asset class covering all properties used by firms. If firms can directly own such properties, they can also rent them from CRE investors. In this paper, I show that taking into account CRE investors enables to better understand the macroeconomic effects of credit supply shocks. I show empirically that CRE investors capture a large portion of firms' value added, are particularly leveraged, and rely heavily on debt to finance their investment. Based on these stylized facts, I then introduce CRE investors in a standard dynamic stochastic general equilibrium model. In this model, I show that credit supply shocks have stark heterogeneous effects between CRE investors and other non-financial firms, and lead to misallocation.

**The growing need for energy flexibility has led to demand-response programs aimed at reducing peak electricity usage. EDF R&D has implemented several initiatives using smart meter data from Linky devices. However, evaluating their effectiveness remains challenging due to individual consumption variability, self-selection bias, and limitations of standard causal inference methods. We simulate a controlled setting to highlight these issues and compare traditional approaches with machine learning techniques such as SyncTwin, which generates synthetic control units to handle unobserved confounders. Our results show that standard methods often fail when time series complexity and hidden variables affect treatment assignment, underscoring the need for more robust causal inference models in energy policy evaluation.