U.S. Equity Risk Premium Forecasting Dataset (1990–2024): Implied ERP, Realized Returns, and Macro-Financial Predictors
Description
This dataset contains a comprehensive quarterly panel of U.S. equity risk premium measures and macro-financial predictors covering the period 1990Q1–2024Q4. It was constructed to support the analysis in “Structural Shifts in the Forecasting Dynamics of the U.S. Equity Risk Premium: A Regime-Switching and Structural Break Approach, 1990–2024.” The dataset includes two principal dependent variables: (1) Implied Equity Risk Premium (ERP_implied) based on Damodaran’s monthly U.S. ERP estimates, sampled at quarter end; (2) Realized 1-Year Equity Risk Premium (ERP_realized_1y) computed as the difference between S&P 500 total returns and 3-month Treasury bill returns over a four-quarter horizon. It also provides the key macro-financial predictors widely used in the forecasting literature, including: – Valuation ratios: Earnings-Price (EP_ratio), Dividend-Price (DP_ratio) – Interest-rate variables: 3-month Treasury bill rate, 10-year Treasury yield, term spread, and Moody’s Baa–Treasury credit spread – Macroeconomic indicators: Real GDP growth, CPI inflation, unemployment rate, and the federal funds rate – Risk and uncertainty indicators: VIX volatility index and the U.S. Economic Policy Uncertainty (EPU) index – Realized volatility: Annualized standard deviation of monthly S&P 500 total returns over the prior 12 months. All variables are aligned to quarterly frequency (end-of-quarter levels for stock-market and ERP series; quarterly averages for flow-type indicators such as VIX, EPU, and macroeconomic rates). The dataset is cleaned, merged, and structured to allow direct replication of forecasting regressions, Bai–Perron structural break tests, and two-state Markov-switching models. This dataset supports empirical research on the time variation, structural instability, and regime dependence of the U.S. equity risk premium. It may be used for forecasting studies, return predictability analysis, macro-financial modeling, and teaching applications in asset pricing and financial economics.
Files
Steps to reproduce
The dataset is provided as a single Excel file (US_ERP_Final_v7.xlsx) containing quarterly observations from 1990Q1 to 2024Q4. Each variable name corresponds directly to the notation used in the accompanying research article. The following steps reproduce the dataset from raw sources: 1. Retrieve raw data – Download monthly implied ERP estimates from Aswath Damodaran’s website (2024 and 2025 ERP datasets). – Obtain S&P 500 index level, monthly total return series, trailing 12-month earnings, and trailing 12-month dividends from public market-data providers (e.g., Yahoo Finance, FRED). – Download the 3-month Treasury bill rate, 10-year Treasury yield, and Moody’s Baa corporate bond yield from the Federal Reserve Bank of St. Louis (FRED). – Obtain macroeconomic variables (GDP, CPI inflation, unemployment rate, federal funds rate) from BEA and BLS/FRED. – Download daily or monthly VIX data from the CBOE website, and the U.S. Economic Policy Uncertainty (EPU) index from the Baker–Bloom–Davis project. 2. Convert raw data to quarterly frequency – Use end-of-quarter values for implied ERP, valuation ratios, yields, and spreads. – Use quarterly averages for VIX, EPU, and short-rate variables. – Align GDP growth, inflation, unemployment, and federal funds rate to quarterly frequency based on official releases. 3. Construct derived variables – Implied ERP: take Damodaran’s monthly ERP series and sample the quarter-end observation. – Realized 1-Year ERP: compute ERP_(realized_(t+4) )= γ + δ1 ERP_(implied_t )+ δ2^' X_t+ u_(t+4) realized =100×(R t,t+4 EQ −R t,t+4 RF ), where equity total return is the 12-month S&P 500 return and the risk-free return is the compounded 3-month T-bill rate. – Valuation ratios: compute earnings-price (EP_ratio) and dividend-price (DP_ratio) from trailing 12-month earnings and dividends divided by the quarter-end index level. – Term spread: yield(10y) minus yield(3m). – Credit spread: Moody’s Baa yield minus 10-year Treasury yield. – Realized volatility: annualized standard deviation of monthly S&P 500 returns over the previous 12 months. – All variables are merged on a consistent quarterly date index. 4. Clean and validate data – Check for missing observations and interpolate only where the source series provides official backward-filled values (e.g., monthly VIX gaps). – Ensure monotonic timestamps, remove duplicates, and confirm unit consistency (percent vs decimal). – Verify alignment by confirming that all variables share identical quarterly timestamps from 1990Q1 to 2024Q4. 5. Export dataset – Save the final merged table as US_ERP_Final_v7.xlsx, with variable names matching those used in the empirical analysis. – The dataset is ready for replication of forecasting regressions, Bai–Perron structural break tests, and two-state Markov-switching models used in the study.
Institutions
- Universita Ca' Foscari