Monetary Debasement and Equity Tail Risk: Scenario Generation via Global/Local Conditioning on Macroeconomic Paths

Unconventional monetary policy, which became routine through the global financial crisis and the COVID-19 pandemic, has sustained debate over the erosion of monetary purchasing power and over asset inflation. Because capital markets record only what actually happened, however, data on tail risk in response to monetary policy and liquidity expansion are limited. For this reason, conditional scenarios that reflect the correlations among variables become a meaningful tool. Yet existing deep generative models in finance have largely concentrated on prediction; they rarely consider hypothetical conditions and, still more rarely, conditional paths. This study proposes MAC-Flow, a model that generates equity price scenarios up to 13 weeks ahead from the trajectories of the 3-month Treasury bill rate and an excess liquidity indicator. MAC-Flow is an autoregressive (AR) conditional normalizing flow that generates scenarios on hypothetical paths through a dual mechanism: a whole-path encoder for global broadcasting and a per-step encoder for pointwise injection. Against the other deep generative models, condGAN and condVAE, MAC-Flow showed a limited advantage on the pooled-fold sign test (p = 0.00024), and it was not superior to GARCH-ST in CRPS. MAC-Flow, however, reflects the time-varying hypothetical path and is therefore suited to measuring tail risk, such as the intra-horizon loss, and in the comparison with the AR-based VAE of the same structure, MAC-Flow produced better skewness (p < 0.0001). This study offers a short-horizon scenario generator that focuses on tail risk linked to monetary policy, and it can serve risk management under shifting policy conditions.