The aim of this article is to determine whether Monetary Condition Indices are useful indicators for future economic activity. First, two versions of MCI are successively studied (Long-term MCIs defined with long-term weights and standard MCIs like those built by the IMF). In-sample regressions, out-of-sample simulations and probit analysis (intended to determine the capacity of MCIs to announce downturns) indicate that the informational content of MCIs is very sporadic. We then try to identify the reasons for these poor results, focusing on the fact that MCIs do not take into account the dynamic characteristics of its components. So, as exchange rate, interest rates and asset prices affect the economic activity with different forces and with delayed responses, it seemed important to consider past evolutions of these variables, with relative weights varying for each lag considered. Proof of the importance of past shocks that are still "in the pipeline", we demonstrate that such a Dynamic-Weight MCI constitutes a better
