Paper: arXiv 2512.20477
Authors: Ilias Aarab
Abstract
In Aarab (2020), I examine U.S. stock return predictability across economic regimes and document evidence of time-varying expected returns across market states in the long run. The analysis introduces a state-switching specification in which the market state is proxied by the slope of the yield curve, and proposes an Aligned Economic Index built from the popular predictors of Welch and Goyal (2008) (augmented with bond and equity premium measures). The Aligned Economic Index under the state-switching model exhibits statistically and economically meaningful in-sample (\(R^2 = 5.9%\)) and out-of-sample (\(R^2_{"\text{oos"}} = 4.12%\)) predictive power across both recessions and expansions, while outperforming a range of widely used predictors. In this work, I examine the added value for professional practitioners by computing the economic gains for a mean-variance investor and find substantial added benefit of using the new index under the state switching model across all market states. The Aligned Economic Index can thus be implemented on a consistent real-time basis. These findings are crucial for both academics and practitioners as expansions are much longer-lived than recessions. Finally, I extend the empirical exercises by incorporating data through September 2020 and document sizable gains from using the Aligned Economic Index, relative to more traditional approaches, during the COVID-19 market turbulence.
Complexity vs Empirical Score
- Math Complexity: 6.0/10
- Empirical Rigor: 7.5/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs a state-switching regression model and partial least squares (PLS) for index construction, indicating non-trivial econometric complexity. It also features extensive empirical validation, including in-sample and out-of-sample metrics, portfolio allocation exercises with transaction costs, and analysis through September 2020 to assess performance during COVID-19 turbulence.
Research Flowchart
flowchart TD
A["Research Goal<br>Predict U.S. Equity Premium across Economic Regimes"] --> B{"Methodology"}
B --> C["State-Switching Model<br>Market State = Yield Curve Slope"]
B --> D["Aligned Economic Index<br>Welch & Goyal (2008) + Bond/Equity Premium"]
C --> E["Computational Process<br>In-Sample & Out-of-Sample Analysis"]
D --> E
E --> F["Outcomes & Findings"]
F --> G["In-Sample: R² = 5.9%<br>Out-of-Sample: R² = 4.12%"]
F --> H["Economic Gains<br>Mean-Variance Investor"]
F --> I["COVID-19 Extension<br>Sizable gains during turbulence"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a state‑switching return‑prediction model that uses an Aligned Economic Index built from popular predictors, aiming to improve stock return predictability across market regimes.
Source sentence
The analysis introduces a state‑switching specification in which the market state is proxied by the slope of the yield curve, and proposes an Aligned Economic Index built from the popular predictors of Welch and Goyal (2008) (augmented with bond and equity premium measures).
Assumes 6
- Investors have mean‑variance preferences and maximize expected utility of terminal wealth. model
Source sentence
Consider an investor with a one‑month horizon who chooses portfolio weights to maximize expected utility of terminal wealth Wt+1 conditional on information available at time t.
- Forecasts are generated recursively using only information available up to the forecast formation date to avoid look‑ahead bias. statistical
Source sentence
re estimated recursively using only information available up to the forecast formation date t, so the procedure avoids look‑ahead bias.
- The investor allocates only between the S&P 500 index and U.S. Treasury bills. market
Source sentence
The investor allocates between the S&P 500 index and U.S. Treasury bills (or a combination of both).
- Transaction costs are modeled as a proportional cost of 50 basis points per transaction. execution
Source sentence
The “cost = 50bps” statistic is the CER gain assuming a proportional transactions cost of 50 basis points per transaction.
Show 2 more
- Historical‑average forecast serves as a benchmark for turnover and CER comparisons. data
Source sentence
Relative average turnover is computed as average turnover divided by the average turnover of the investor who uses the historical‑average forecast.
- The post‑war sample (January 1950‑September 2020) is representative for evaluating predictability. data
Source sentence
I use the updated dataset of Welch and Goyal (2008) covering 14 widely used fundamental predictors from January 1950 to September 2020, supplemented by bond- and equity‑premium measures.
Data
1980-01 – 2020-09 · S&P 500 stocks · equities · monthly
- Welch and Goyal (2008) updated dataset fundamentals public
Source sentence
The data can be retrieved from Amit Goyal’s web page at http://www.hec.unil.ch/agoyal/. A detailed description is provided in Welch and Goyal (2008).
- S&P 500 excess return series other public
Source sentence
The aggregate stock market return is measured as the excess return: the continuously compounded log return on the S&P 500 index (including dividends) minus the risk‑free rate (proxied by the three‑month Treasury bill).
- Yield‑curve spread indicator macro public
Source sentence
…dicator can be retrieved from the Federal Reserve Bank of St. Louis.
Costs
Modeled
A proportional cost of 50 basis points per transaction is applied when reporting net CER gains.
Source sentence
The “cost = 50bps” statistic is the CER gain assuming a proportional transactions cost of 50 basis points per transaction.
Finds 5
- The Aligned Economic Index under the state‑switching model achieves in‑sample R² of 5.9% and out‑of‑sample R²ₒₛ of 4.12% across recessions and expansions. other
Source sentence
The Aligned Economic Index under the state‑switching model exhibits statistically and economically meaningful in‑sample ($R^2 = 5.9\%$) and out‑of‑sample ($R^2_{\text{oos}} = 4.12\%$) predictive power across both recessions and expansions.
- Forecasting models (EPLS, EPCA, EFC) deliver higher Sharpe ratios (0.16) than the historical‑average benchmark (0.13). performance
Source sentence
the Sharpe ratios of the forecasting models all exceed that of the historical average (0.13), with EPLS and EFC yielding the highest ratio of 0.16.
- After accounting for transaction costs, only the state‑switching Aligned Economic Index provides statistically significant certainty‑equivalent return (CER) gains across market states. performance
Source sentence
I find that using EPLS within the state‑switching framework delivers economically large and statistically significant certainty‑equivalent return gains for a mean‑variance investor across market states, and outperforms all considered alternatives, including buy‑and‑hold.
- Historical‑average forecast yields large CER gains in expansions (13.00%) and up states (10.74%) but negative gains in recessions (‑17.46%) and down states (‑1.08%). performance
Source sentence
When looking across different states, the historical‑average model produces substantial ∆CER gains during expansions and up states of 13.00% and 10.74%, respectively, but yields negative gains across recessions and down states of -17.46% and -1.08%.
- During the COVID‑19 episode, the state‑switching Aligned Economic Index continues to generate positive CER gains, unlike many traditional predictors whose gains are concentrated in contractions. performance
Source sentence
These findings matter for practitioners because expansions are long‑lived relative to recessions: unlike many predictors whose gains are concentrated in contractions, the state‑switching Aligned Economic Index delivers value that is more robust across the business cycle and can be implemented on a consistent real‑time
Checked by others
Supported by
- Stock Price Predictability and the Business Cycle via Machine Learning
Compare the findings
This paper: Historical‑average forecast yields large CER gains in expansions (13.00%) and up states (10.74%) but negative gains in recessions (‑17.46%) and down states (‑1.08%).
That paper: ML models exhibit higher mean‑squared error during recessions than expansions, confirming that recessions degrade forecasting accuracy.
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (40)
- A Comprehensive Look at the Empirical Performance of Equity Premium Prediction (2008) — link
- Dividend Yields and Expected Stock Returns (1988) — link
- Stock Prices, Earnings, and Expected Dividends (1988)
- On the Predictability of Stock Returns: An Asset-Allocation Perspective (1996) — link
- On the Out-of-Sample Predictability of Stock Market Returns (2002) — link
- Predicting Returns with Financial Ratios (2004) — link
- Cross-Sectional Forecasts of the Equity Premium (2006)
- Forecasting Stock Returns: A Predictor-Constrained Approach (2018)
and 32 more
- Investor Sentiment Aligned: A Powerful Predictor of Stock Returns (2015) — link
- Regime Shifts and Return Predictability (2018)
- Manager Sentiment and Stock Returns (2019) — link
- Why Does Return Predictability Concentrate in Bad Times? (2017) — link
- Is Stock Return Predictability Time-Varying? (2018) — link
- Path Models with Latent Variables: The NIPALS Approach (1975) — link
- Market Expectations in the Cross Section of Present Values (2013) — link
- The Three-Pass Regression Filter: A New Approach to Forecasting Using Many Predictors (2015) — link
- Out-of-Sample Equity Premium Prediction: Combination Forecasts and Links to the Real Economy (2010) — link
- Presidential Address: Discount Rates (2011) — link
- International Stock Return Predictability: What Is the Role of the Real Economy? (2016)
- Predicting Excess Stock Returns out of Sample: Can Anything Beat the Historical Average? (2008)
- Forecasting the Equity Risk Premium: The Role of Technical Indicators (2014) — link
- Portfolio Selection (1952)
- Return Predictability Revisited Using Weighted Least Squares (2017)
- Time-Consistent Portfolio Selection under Short-Selling Prohibition: From Discrete to Continuous Setting (2014) — link
- Practical Applications of Return Predictability and Dynamic Asset Allocation: How Often Should Investors Rebalance? (2015)
- Asymptotic Inference for Performance Fees and the Predictability of Asset Returns (2018) — link
- Market Timing over the Business Cycle (2018) — link
- Time-Varying Short-Horizon Predictability (2011) — link
- The Evolution of Corporate Dividend Policy (1997)
- Approximately Normal Tests for Equal Predictive Accuracy in Nested Models (2007) — link
- The Stock Market’s Reaction to Unemployment News: Why Bad News Is Usually Good for Stocks (2005) — link
- Predictive Regressions with Time-Varying Coefficients (2012) — link
- In Search of Attention (2011) — link
- Economic Implications of Bull and Bear Regimes in UK Stock and Bond Returns (2005) — link
- Business Conditions and Expected Returns on Stocks and Bonds (1989) — link
- Predictive Regressions (1999) — link
- Constraints Forecasting Stock Returns under Economic Constraints (2014)
- The Yield Curve and Predicting Recessions (2006) — link
- Learning about Predictability: The Effects of Parameter Uncertainty on Dynamic Asset Allocation (2001) — link
- The Aligned Economic Index & the State-Switching Model (2020) — link
0 of 40 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.