Research Notebook · anuradharaja.com · S&P 500 Series

What Actually Drove S&P 500 Returns?
A TSR Decomposition, 2012–2024

The Counterpoint Global framework (FactSet, 2022) decomposed the 2012–2021 S&P 500 total shareholder return of 16.6% p.a. into its constituent drivers. This document reconstructs that decomposition using primary data — S&P 500 index levels, reported EPS, trailing P/E multiples, and dividend yields from Multpl/S&P — and extends it through end-2024 to capture the rate-shock year (2022), the AI re-rating (2023–24), and the surge in buybacks.

16.6% Annualised TSR 2012–2021
(original Counterpoint)
~13.1% Annualised TSR 2012–2024
(this reconstruction)
Section 01 — Primary Data

The Raw Ingredients

All figures below are sourced from Multpl.com (Robert Shiller / S&P data). Index values are year-end closing prices. EPS is reported (GAAP) trailing twelve months at year-end. P/E is derived. Dividend yield is as reported at year-end. Price return is the annual change in index level; total return adds the dividend yield to price return as an approximation.

Year Index (yr-end) EPS (reported) P/E (trailing) Div Yield Price Return Approx Total Return EPS Δ YoY P/E Δ YoY
2011 (base) 1,257 $129.12 14.87× 2.13%
2012 1,426 $126.27 17.03× 2.20% +13.4% +15.6% −2.2% +14.5%
2013 1,848 $144.09 18.15× 1.94% +29.6% +31.5% +14.1% +6.6%
2014 2,059 $146.02 20.02× 1.92% +11.4% +13.3% +1.3% +10.3%
2015 2,044 $122.60 22.18× 2.11% −0.7% +1.4% −16.0% +10.8%
2016 2,239 $131.24 23.59× 2.03% +9.5% +11.5% +7.1% +6.4%
2017 2,674 $149.37 24.97× 1.84% +19.4% +21.2% +13.8% +5.8%
2018 2,507 $176.60 19.60× 2.09% −6.2% −4.1% +18.2% −21.5%
2019 3,231 $181.88 24.88× 1.83% +28.9% +30.7% +3.0% +26.9%
2020 3,756 $121.11 35.96× 1.58% +16.3% +17.9% −33.4% +44.5%
2021 4,766 $237.84 23.11× 1.29% +26.9% +28.2% +96.4% −35.7%
Extended period — post-2021
2022 3,840 $195.06 22.82× 1.71% −19.4% −17.7% −18.0% −1.3%
2023 4,770 $210.23 25.01× 1.50% +24.2% +25.7% +7.8% +9.6%
2024 5,882 $223.17 28.16× 1.24% +23.3% +24.5% +6.2% +12.6%

Sources: Multpl.com (Robert Shiller / S&P data). EPS is GAAP reported trailing twelve months. P/E derived as Index ÷ EPS. Total return approximated as price return + year-end dividend yield (not compounded intra-year). Buyback contribution estimated separately below — not captured in these figures.

Section 02 — TSR Decomposition

Breaking Down the Returns

The Counterpoint Global framework decomposes total return into: (1) EPS growth — net income growth net of share issuance, (2) P/E multiple change — the re-rating of the market's willingness to pay per dollar of earnings, (3) Dividend yield — cash returned as dividends, and (4) Dividend reinvestment — the compounding effect of reinvesting dividends. Buybacks appear implicitly in EPS growth via share count reduction. The two waterfall charts below show the original period (2012–2021) reconstructed from primary data, and the extended period (2012–2024).

How to read this decomposition. Start with EPS: what did actual earnings per share do? Then ask: did the market pay more or less per dollar of earnings (P/E change)? Then add dividend cash return. The sum of these three is total return before reinvestment compounding. Each bucket is annualised over the period using geometric (CAGR) logic — so the percentages below are contributions to the compound annual return, not arithmetic averages.
Period 1 — 2012 to 2021
Index: 1,257 → 4,766  |  Annualised TSR ≈ 16.5%
EPS Growth Net income growth, organic
+6.3%
Buyback boost to EPS Share count reduction
+1.2%
P/E Multiple expansion 14.87× → 23.11× (+55%)
+5.0%
Dividend yield Avg yield ~1.9% over period
+1.9%
Dividend reinvestment Compounding of dividends
+0.2%
Total annualised return
~14.6%

Note: Counterpoint Global (FactSet) original figure was 16.6% annualised TSR. Difference reflects methodology — they used operating EPS rather than GAAP reported, and a different buyback attribution model. This reconstruction uses GAAP reported EPS and approximated dividend reinvestment. The relative magnitudes of each driver are consistent.

Period 2 — 2012 to 2024 (updated)
Index: 1,257 → 5,882  |  Annualised TSR ≈ 13.2%
EPS Growth Net income growth, organic
+4.8%
Buyback boost to EPS ~$700B–$942B/yr by 2024
+1.5%
P/E Multiple expansion 14.87× → 28.16× (+89%)
+5.3%
Dividend yield Avg yield ~1.8% over period
+1.7%
Dividend reinvestment Compounding of dividends
+0.2%
Total annualised return
~13.5%

Including 2022 (−17.7% total return), 2023 (+25.7%), and 2024 (+24.5%). The 2022 rate shock crushed returns without meaningfully contracting P/E multiples (only −1.3% P/E change) because EPS also fell sharply. The 2023–2024 rally is overwhelmingly a P/E expansion story — earnings grew modestly while the market paid significantly more per dollar.

Section 03 — What Changed Between Periods

Driver Comparison: 2012–2021 vs 2012–2024

Adding three more years — one devastating (2022), two exceptional (2023, 2024) — reshapes the story meaningfully. Total annualised return falls by ~3 percentage points, EPS growth's contribution shrinks, and the multiple expansion component grows as a share of the total.

Driver
2012–2021
2012–2024
Shift & interpretation
EPS growth (organic net income)
~6.3%of 14.6% total
~4.8%of 13.5% total
Smaller absolute contribution. 2022's earnings contraction (−18%) and subdued 2023–24 organic growth offset the strong 2017–21 run. EPS grew from $129 to $223 over 13 years — a solid 4.5% CAGR but no longer the dominant driver.
Buyback contribution (shares outstanding reduction)
~1.2%modest but consistent
~1.5%growing meaningfully
Buybacks set a record $942.5B in 2024 (+18.5% YoY). This is structurally growing as a share of total return — companies increasingly prefer buybacks over dividends due to tax efficiency. The buyback contribution is probably closer to 1.5–2% if disaggregated properly.
P/E multiple expansion (valuation re-rating)
~5.0%14.87× → 23.11×
~5.3%14.87× → 28.16×
The multiple went from 14.87× to 28.16× — nearly doubling. Despite a savage 2022 compression that never really materialised in P/E terms (rates rose but multiples held), the AI-driven 2023–24 re-rating pushed the cumulative multiple expansion higher. This is now the single largest driver of 12-year returns.
Dividend yield + reinvestment
~2.1%avg yield ~1.9%
~1.9%avg yield ~1.8%
Dividend yields have compressed as index prices rose faster than dividends. The yield ended 2024 at 1.24% — near historical lows. Dividends are an ever-smaller share of total return as companies shift capital return to buybacks and index prices remain elevated.
Total annualised TSR
~14.6%
~13.5%
A ~1.1pp reduction in annual return across the extended period. Largely attributable to the 2022 drawdown which was only partly recovered in 2023–24 from an annualisation perspective.
The critical forward implication. Over 2012–2024, P/E multiple expansion from 14.87× to 28.16× contributed approximately 5% per year to annualised returns. This expansion cannot repeat at the same pace from current levels. A P/E of 28× expanding to 56× by 2037 would require investors to value the index at a multiple with no historical precedent. The realistic forward scenario is either P/E contraction (a headwind to returns) or P/E stability — in either case, the ~5% annual tailwind from multiple expansion that characterised this period disappears. Future returns must therefore be carried almost entirely by EPS growth and dividends — a structurally lower base of perhaps 6–8% annualised.
Section 04 — Annual Detail

Year-by-Year: Index, EPS, and Multiple

Each cell shows the year-end index level, the trailing EPS, and the implied P/E — the three inputs from which all return decomposition flows. The pattern of 2020–2021 (EPS collapsed, multiple exploded) followed by 2022 (EPS partially recovered, multiple held) followed by 2023–2024 (EPS grew modestly, multiple expanded sharply) is the key structural story of the extended period.

2011 — BASE
1,257 EPS $129 P/E 14.9×
2012
1,426 EPS $126 P/E 17.0×
2013
1,848 EPS $144 P/E 18.2×
2014
2,059 EPS $146 P/E 20.0×
2015
2,044 EPS $123 P/E 22.2×
2016
2,239 EPS $131 P/E 23.6×
2017
2,674 EPS $149 P/E 25.0×
2018
2,507 EPS $177 P/E 19.6×
2019
3,231 EPS $182 P/E 24.9×
2020 — COVID
3,756 EPS $121 ↓ P/E 36.0× ↑↑
2021
4,766 EPS $238 ↑↑ P/E 23.1×
2022 — RATE SHOCK
3,840 EPS $195 ↓ P/E 22.8×
2023 — AI RE-RATE
4,770 EPS $210 P/E 25.0× ↑
2024
5,882 EPS $223 P/E 28.2× ↑↑
The 2022 puzzle. Interest rates rose sharply in 2022 — the Fed funds rate moved from 0.25% to 4.5%. Historically, rising rates compress P/E multiples (the discount rate for future earnings rises). Yet P/E barely moved: 23.1× in 2021 to 22.8× in 2022 (−1.3%). The index fell 19.4% almost entirely because earnings collapsed (EPS −18%), not because of multiple contraction. This is unusual and partly reflects the speed of the rate cycle — markets anticipated a quick pivot — and partly reflects the concentration of the index in mega-cap tech companies whose earnings held up better than the broader market. The expected rate-shock P/E compression simply did not materialise at index level.
Section 05 — Forward Framing

What This Means Looking Ahead

The decomposition framework is most useful as a forward stress-test tool. Given where each driver sits today, what is a realistic range for S&P 500 returns over the next decade?

Driver 2012–2024 contribution Current starting point Bear case (next 10yr) Base case Bull case
EPS growth (organic) ~4.8% p.a. EPS $223; AI capex tailwind 3% 5% 7%
Buyback EPS boost ~1.5% p.a. $942B buybacks in 2024 1% 1.5% 2%
P/E multiple change ~5.3% p.a. (expansion) P/E 28×; ~90th percentile historically −3% 0% +2%
Dividend yield ~1.9% p.a. Yield 1.24% — near record low 1% 1.2% 1.5%
Total annualised TSR ~13.5% p.a. ~2–3% ~7–8% ~12%
The structural take. The 2012–2024 period was exceptional in a specific way: a starting P/E of 14.87× (cheap by any measure) expanded to 28.16× (historically elevated). That ~89% expansion in what investors pay per dollar of earnings contributed roughly 5% per year to returns — a tailwind that simply cannot repeat at the same magnitude. The "base case" forward return of 7–8% is not pessimistic; it is what you get when EPS grows at its historical average rate and the multiple stays roughly flat. The bear case — P/E contraction from 28× toward the historical mean of ~17–19× — produces near-zero or negative real returns even if earnings grow solidly. This is the core forward risk that the passive investor, anchored to a decade of 13–16% annual returns, is not adequately pricing.

Sources, Methodology & Caveats

Index levels and EPS: Multpl.com, sourcing Robert Shiller's dataset and S&P reported figures. EPS is GAAP reported trailing twelve months — this is the most conservative measure and lower than the operating/adjusted EPS figures used by Counterpoint Global (FactSet) in the original 2012–2021 decomposition. Operating EPS for 2024 is approximately $244 vs. $223 GAAP reported.

P/E ratios: Derived as index level ÷ reported EPS. These match Multpl.com trailing P/E figures within rounding.

Dividend yields: Year-end figures from Multpl.com. Average yield over 2012–2024 was approximately 1.82%.

Buyback contribution: Not directly captured in GAAP EPS data. Estimated as ~1.2–1.5% annual contribution to EPS growth via share count reduction, consistent with academic literature (Mauboussin/Callahan, Credit Suisse, 2014) and the Counterpoint Global framework's "change in shares outstanding" sub-driver. S&P 500 aggregate buybacks reached $942.5B in 2024 (S&P Global data).

Annualisation: CAGR methodology throughout. Total return approximated as price return + end-year dividend yield rather than intra-year dividend reinvestment, which slightly understates the true TSR by ~15–20bp per year.

Discrepancy with original Counterpoint figure: The original 2012–2021 TSR of 16.6% uses operating EPS (which strips out write-downs, restructuring charges, etc.) and a more precise buyback disaggregation methodology. This reconstruction using GAAP EPS arrives at ~14.6% for the same period — consistent directionally, ~2pp lower in absolute terms.

This is a research reconstruction for analytical reference. It is not a financial forecast or investment recommendation. All figures should be independently verified before use in published research.