Beyond Headline Inflation: A Components Approach to Risk Premia with PCA Extensions

This dissertation investigates how inflation risks are priced across asset classes by decomposing headline inflation and extending the analysis through principal components analysis (PCA). The goal is to understand the distinct economic channels through which inflation affects asset returns and to develop more precise inflation hedging strategies for diversified portfolios.

The first paper, What Lies Beneath the Headline: Analyzing Inflation Risks Within and Across Asset Classes, examines how core and energy components of inflation affect returns across stocks, bonds, commodities, and currencies from 1967 to 2022. Using monthly data in a Fama-MacBeth regression framework, the paper shows that headline inflation generally produces negative inflation betas for stocks and bonds, while commodities exhibit positive betas. However, decomposing inflation into its core and energy components reveals distinct dynamics: core inflation yields negative betas and statistically insignificant risk premia, whereas energy inflation is associated with positive betas and a statistically significant positive risk premium. These findings underscore the importance of separating inflation components when analyzing asset pricing, as headline inflation masks the contrasting effects of core and energy shocks.

The second paper, Extracting Latent Signals in Headline Inflation: A Principal Components Approach to Risk Premia, builds on this decomposition by introducing a data-driven framework for identifying hidden inflation risks. It applies PCA to granular Consumer Price Index (CPI) subcomponents to derive new inflation factors that capture dimensions not observable in core and energy measures alone. A novel factor—PC3ResCE—is constructed by residualizing the third principal component with respect to core and energy inflation innovations. This factor is statistically significant and economically interpretable, capturing pro-cyclical inflation dynamics tied to discretionary consumption and economic growth. It improves model fit substantially, increasing explanatory power from 20.5% to 47.7% in the Fama-MacBeth regressions. A second derived factor, PC9ResCE, is also identified and shown to be jointly significant when included alongside PC3ResCE, core, and energy inflation. Together, these PCA-derived factors offer a more complete picture of inflation risk premia and highlight the value of using both economic intuition and statistical decomposition to model inflation's impact on asset prices.

A 22-page response to the dissertation defense is included in the appendix. It addresses methodological concerns related to the use of contemporaneous versus lagged inflation data, revised versus vintage data, and the economic interpretation of statistical factors. Robustness checks confirm that key findings—particularly the significance of energy inflation and residual PCA factors—are stable across data vintages and specifications. The response also expands the analysis to include multivariate regressions with PC9ResCE, demonstrating that inflation risk premia likely reflect multiple dimensions of macroeconomic risk.

Together, the papers and supporting response contribute to the literature by establishing that inflation is not a monolithic risk. Instead, it comprises diverse components—some persistent, others cyclical—that are priced differently across asset classes. By combining economic decomposition and statistical factor extraction, this research enhances both theoretical understanding and practical tools for inflation risk management.