Impact of macroeconomic shocks on extreme stock returns: a quantile approach

Details

Supervisors

Iania, Leonardo

Faculty

Louvain School of Management

Degree label

Master [120] : ingénieur de gestion, à finalité spécialisée

Abstract

enThis thesis investigates how key macro-financial variables, including market volatility (VIX), credit spreads (BAA–10Y) and interest rate changes (10Y–3M), affect stock returns across six major equity indices: the S&P 500, NASDAQ 100, CAC 40, DAX 30, Nikkei 225 and Hang Seng. Covering the period from January 2000 to January 2024, the analysis uses a variety of econometric methods such as Ordinary Least Squares (OLS), Quantile Regression (QR), Quantile Generalized Additive Models (QGAM) and Quantile Regression Forests (QRF) to explore how these relationships vary across different market environments and segments of the return distribution. The results show that the effects of macro-financial variables on returns vary depending on market conditions. Volatility, as measured by the VIX, has a consistently negative impact, particularly in the lower quantiles, which suggests that uncertainty affects returns most strongly during periods of market stress. Credit spreads display a more nuanced pattern, showing a positive relationship with returns during bearish conditions. This counterintuitive finding may reflect a shift toward riskier assets in down markets, although it might also result from model limitations or unobserved factors. Interest rate changes, on the other hand, have weak and inconsistent effects, with no clear link to stock returns. By comparing the four models, the study highlights the value of quantile-based and nonlinear approaches in capturing the asymmetric and context-specific behaviour of financial markets. These findings provide valuable insights for investors who seek to manage risk under different market conditions, but also for policymakers who monitor volatility and credit dynamics to support financial stability. The thesis adds to the literature on the predictability of returns and highlights the limitations of traditional mean-based models.