S&P 500 Monthly Data Since 1871 – Shiller/datahub.io
- Rows
- 1,869
- Columns
- 10
- Data range
- 1871-01-01 to 2026-09-01
- Auto-refresh
- Re-checks its source every 14 days
- Last refreshed with new data
- 2026-10-12
Monthly S&P 500 index data since January 1871. Includes SP500 price, Dividend, Earnings, CPI, Long Interest Rate, Real Price, Real Dividend, Real Earnings, and PE10 (Shiller CAPE ratio). Extended with FRED data post-2023.
AI analysis
S&P 500 Monthly Data (Shiller/datahub.io) — Analytical Overview
1. Dataset Representation & Correlation Value
This dataset represents one of the most historically rich equity market time series publicly available, spanning 155 years of monthly S&P 500 observations from January 1871 through May 2026 (1,865 rows across 10 columns), sourced from Robert Shiller's foundational macroeconomic research and extended post-2023 via FRED (Federal Reserve Economic Data). The source CSV at https://raw.githubusercontent.com/datasets/s-and-p-500/main/data/data.csv reflects a community-maintained, actively updated pipeline, which adds meaningful freshness credibility relative to static academic exports. The combination of nominal and inflation-adjusted variants of price, dividends, and earnings — alongside CPI and the Shiller CAPE (PE10) — makes this dataset exceptionally well-suited for long-horizon macroeconomic correlation studies, cyclically adjusted valuation research, and real-return modeling. Few publicly available datasets offer this breadth of simultaneously collected financial and price-level variables at monthly granularity across multiple economic regimes.
---
2. Data Quality Observations
Data quality here is notably strong for a dataset of this age and complexity. Zero null cells are recorded across all 10 columns and all 1,865 rows — a remarkable result given that several columns (Earnings, Dividend) represent estimates for 19th-century periods where underlying data was sparse and likely interpolated by Shiller's methodology. Duplicate row counts are flagged as pending Phase D recomputation, so that figure should be validated before any join or aggregation work proceeds; given the monthly date granularity with 1,865 distinct Date values matching exactly 1,865 total rows, duplicates are unlikely but not yet confirmed absent. One subtle quality concern worth flagging: minimum values of 0.0 appear across Dividend, Earnings, CPI, Long Interest Rate, Real Price, Real Dividend, and Real Earnings — in a dataset where true zeros would be economically implausible for most of these fields (e.g., a zero CPI or zero long interest rate), these likely represent placeholder or missing-value encodings rather than genuine observations. Any correlation or regression analysis should filter or impute these zero-floor records prior to modeling, as they could distort distributional assumptions significantly.
---
3. Key Column Distributions & Statistical Highlights
SP500 is the most visually dramatic column: a mean of 471.26 against a median of only 19.83 — a ratio exceeding 23:1 — immediately signals extreme right skew (skewness = 3.38) driven by the exponential price appreciation of the modern era. With Q1 at 8.04 and Q3 at 263.70, the interquartile range captures the pre-1990s regime, while 361 statistical outliers represent largely the post-2000 and especially post-2010 bull market periods. This is not noise; it is regime change, and any correlation model using raw SP500 prices should apply log transformation or use returns instead. PE10 (Shiller CAPE) stands out as the most well-behaved column: skewness of just 0.339, mean of 15.98 nearly matching the median of 15.76, Q1/Q3 of 11.03/20.55, and only 47 outliers — making it the most statistically tractable variable for direct correlation analysis and the one most likely to satisfy normality assumptions without transformation. Real Earnings (mean 40.40, median 26.01, skew 1.95, 183 outliers) and Real Dividend (mean 19.87, median 16.07, skew 1.82) are considerably more symmetric than their nominal counterparts, reinforcing the value of the inflation-adjusted columns for cross-era comparisons. Long Interest Rate (mean 4.41, median 3.77, max 15.32, 190 outliers) reflects the dramatic 1980s rate spike and subsequent secular decline — a key variable for regime-aware analysis.
---
4. Recommended Join Key Columns
The Date column is the unambiguous primary join key: fully populated (nulls = 0), entirely distinct (1,865 unique values matching row count), and formatted as ISO 8601 timestamps at monthly resolution (YYYY-MM-01T00:00:00). For cross-dataset joins, this should be normalized to a year-month key (e.g., YYYY-MM) to accommodate datasets that use different day-of-month conventions for monthly data. A derived Year integer key would additionally support annual-frequency datasets such as GDP, federal budget figures, or annual corporate earnings aggregates. No secondary natural keys (e.g., ticker symbols) are present, which is appropriate given this is an index-level, not security-level, dataset.
---
5. Recommended Pairing Datasets for Correlation Discovery
Several dataset categories would pair powerfully with this data. U.S. GDP and recession indicators (e.g., FRED's NBER recession dates or quarterly real GDP) would enable testing of the well-documented but nuanced relationship between economic output and equity valuations — particularly interesting given the CAPE ratio's forward-return predictive claims. Federal Funds Rate or Treasury yield curve data (also available via FRED) would complement the existing Long Interest Rate column and allow spread-based analysis (e.g., yield curve inversions vs. subsequent SP500 drawdowns). U.S. Corporate Earnings or EPS data by sector from sources like Compustat or Damodaran's annual datasets would allow decomposition of the aggregate Earnings column into sector contributors. Gold prices or commodity indices (e.g., WTI crude, GSCI) would support inflation-hedge and risk-appetite correlation studies, particularly relevant given the CPI column already present. Finally, global equity indices (MSCI World, DAX, Nikkei monthly series) would enable correlation structure analysis across international markets over shared time windows, testing decoupling or contagion hypotheses across the dataset's 155-year span.
Columns
- Date (date)
- SP500 (decimal)
- Dividend (decimal)
- Earnings (decimal)
- Consumer Price Index (decimal)
- Long Interest Rate (decimal)
- Real Price (decimal)
- Real Dividend (decimal)
- Real Earnings (decimal)
- PE10 (decimal)