New study reveals how well stock return forecasts track reality in extreme economic times
image: We compare the performance of ERP_Cs, ERP_Fs, ERP_FCs, and ERP_Zs. The patterns are clear. The winning ERPs are ERP_Cs and ERP_FCs. Both have obviously lower measures of time-series and cross-sectional MEVs.
Credit: Huan Yang (Sichuan Normal University, China) Jun Cai (City University of Hong Kong, Hong Kong) Robert Webb (University of Virginia, Charlottesville, USA)
Background and Motivation
Economic conditions and stock market cycles are constantly changing, presenting both risks and opportunities for investors. While many models exist to predict stock returns, few have been tested under extreme economic environments—such as recessions or market bubbles—when accurate forecasts are most critical. China Finance Review International (CFRI) brings you an article titled “Changing Economic Environment and Expected Return Proxies”, which investigates how well different models of expected stock returns perform during the best and worst of times.
Methodology and Scope
The researchers constructed four types of expected return proxies (ERPs): Characteristic-based (ERP_C) – using firm-specific data, Standard risk-factor-based (ERP_F) – with fixed betas, Risk-factor-based with characteristic-varying betas (ERP_FC), Macroeconomic-variable-based (ERP_Z). They evaluated these models using U.S. stock data from August 1960 to December 2022, covering 749 months. The study used six indicators of economic conditions—including leading economic indices, consumer and business confidence, Shiller’s P/E ratio, dividend yield, and market excess returns—to identify extreme periods. Performance was assessed using measurement error variance (MEV) and regression-based comparisons between expected and realised returns.
Key Findings and Contributions
- Best Performing Models: ERP_FC (risk-factor-based with varying betas) and ERP_C (characteristic-based) outperformed others, highlighting the importance of allowing betas to change with firm characteristics.
- Business Cycle Performance: All ERP types tracked realised returns well during extreme business conditions (e.g., low consumer confidence).
- Market Cycle Shortcomings: During extreme stock market highs or lows, expected returns adjusted sluggishly—only capturing about half the actual return movement. In booming markets, some models even predicted negative returns when actual returns were strongly positive.
- Size and Style Matter: Small and value stocks showed the slowest adjustment in expected returns during favourable economic phases.
Why It Matters
This study is among the first to systematically test how well expected return models perform during economic extremes. It reveals a critical gap: while models are reliable in normal or business-cycle-driven conditions, they fall short during market euphoria or panic. This challenges the validity of rational equilibrium models and underscores the need for behavioural finance-integrated approaches that better capture investor psychology during volatile periods.
Practical Applications
- Investors and Fund Managers: They should be cautious when using standard models during market bubbles or crashes. Combining multiple ERP types—especially those incorporating firm characteristics and dynamic risk factors—may improve robustness.
- Financial Analysts and Researchers: They can use these findings to develop next-generation ERPs that incorporate extrapolative behaviour or sentiment-driven adjustments.
- Risk Management: Institutions can better calibrate their risk models by recognising the limitations of existing ERPs during market extremes.
Discover high-quality academic insights in finance from this article published in China Finance Review International. Click the DOI below to read the full-text original! Open access for a limited time!