Macro ReviewHow to read
Outlook · Synthesis · Earnings outlook

The index's own statements put trailing earnings near $311 by Q2 2028

What do the index's own statements, carried forward, say its earnings will be?

Filings to Q2 2026 · holdings of 30 Jun 2026GDP nowcast of 17 Nov 2026 · the Committee's projections of 16 Sept 2026Updated 17 Nov 2026
Trailing EPS, Q2 2028 (the base)$311.02$229.2110th–90th of the draws$409.52from $292.58 in Q2 2026 · 80% band
EPS growth, calendar 2027 (the base)−4.1%statistical leg +4.7% · Q2 2026 start
Pretax margin, filtered18.2%filed 21.4% · mean since 2017 13.3% · Q2 2026
Payout, terminal year78.7%62.9% in year 1 · 1 − g_L ÷ ROE*
Four-quarter error, since 202110.4 · last year's growth 30.9 · 17 originsbeats both benchmarks
The ten's share of index earnings, year 1025.5%35.6% in year 1 · what the index path assumes

In brief

The index's earnings path from its own statements, what moves it, and the record that says how far to trust it
  • Carried forward, the index's own statements put trailing earnings at $311 per index unit in Q2 2028, against $293 in Q2 2026.→ Exhibit 1
  • The pretax margin, filtered to 18.2%, reverts toward its 2017–26 mean of 13.3%, holding earnings flat through 2028.→ Exhibit 3
  • Payout rises from 63% of earnings to 79% by year ten, as growth slows to the .→ Exhibit 4
  • The index path assumes the ten largest companies' share of index earnings falls from 36% to 25% over ten years.→ Exhibit 7
  • Since 2021 the model's four-quarter error averaged 10.4 , against 30.9 for last year's growth carried forward.→ Exhibit 8
What would change this view
What would change this viewTriggerTodayDecided by
Trailing earnings for Q3 2026 leave the one-quarter 80% bandbelow $296.31 or above $321.87$292.58 · Q2 2026pendingthe members' Q3 2026 filings (10-Qs due 9 Nov 2026)
Trailing earnings for Q2 2027 leave the four-quarter 80% bandbelow $262.09 or above $369.71$292.58 · Q2 2026pendingthe members' Q2 2027 filings
The margin stops revertingthe filtered margin above 18.2% a year on18.2% · Q2 2026pendingthe Q2 2027 filings
The reasoning behind each point
  • Path
    Carried forward, the index's own statements put trailing earnings at $311 per index unit in Q2 2028, against $293 in Q2 2026.

    The base runs the S&P 500's income statement, payout and balance sheet on SPY's basis for eight quarters from Q2 2026: revenue from nominal GDP (the GDP nowcast for Q3 2026, then the Committee's medians), the dollar and oil held at their latest months, and a pretax margin that reverts toward its mean. Four in five draws end between $229 and $410. Calendar 2026 comes to $322.84 (+34.3%) and 2027 to $309.57 (−4.1%). Model-conditional: what the statements imply if the model is right, not a forecast of what will be filed.

  • Margin
    The pretax margin, filtered to 18.2%, reverts toward its 2017–26 mean of 13.3%, holding earnings flat through 2028.

    One quarter's filings put the margin at 21.4% in Q2 2026; the margin equation, which separates the filings' one-quarter disturbances (write-downs, one-offs; 10.6 log points a quarter) from the margin itself, reads 18.2%. It closes 16% of the gap to its mean each quarter, and revenue growth lifts it (0.60 per point). The mean is the sample's; at the 2017–19 mean year-2 earnings are $286, at the mean since 2021 $324 — which regime it reverts to is the open question.

  • Payout
    Payout rises from 63% of earnings to 79% by year ten, as growth slows to the long-run rate.

    Dividends follow Lintner's rule (adjusting 2.6% of the way to a 61% target each quarter, estimated on Shiller's series since 1990) and buybacks a ratio that starts at today's 39% of earnings. From year 3 return on equity fades from 18.9% to its long-run 18.8% (opening book), and the payout is what that leaves: at the terminal year 1 − 4.0% ÷ 18.8% = 78.7% net of issuance and the equity residual, 84.1% gross. Shares fall 1.87% a year net, so per-share growth runs at 6.18% against 4.0% for the aggregate.

  • Ten
    The index path assumes the ten largest companies' share of index earnings falls from 36% to 25% over ten years.

    Run on their own levels with the index's dynamics, the ten largest by the fund's value on 30 Jun 2026 (Apple Inc, Nvidia Corp, Microsoft Corp and seven more; 37.9% of the fund) have a mean margin of 28.4% and the rest 11.8%. The rest on its own equation leaves the ten what the index path does not: their implied margin falls to 21.1% by year 10. The two buckets' own sum runs 9.2% above the index path by then — the aggregation effect.

  • Record
    Since 2021 the model's four-quarter error averaged 10.4 log points, against 30.9 for last year's growth carried forward.

    Started at each quarter from the filings as first reported and run on the macro path as it came (17 origins at four quarters, 13 at eight — thin): one quarter ahead 2.8 against 6.4 and 5.1 for Shiller's trend; eight quarters ahead 15.9 against 13.1 for the trend. The band, widened by the factors the margin equation needed on the national accounts since 1970, held 88% at four quarters (59% before widening). The parameters are today's fit: the data are point in time, the parameters are not.

  • Growth fan
    The default model puts 2027 EPS growth at +4.7%; refitted each year since 2005, it did no better than a fixed mean.

    Model-conditional, not a forecast: the growth process fitted to national-accounts profits after tax, run forward 20,000 times from Q2 2026 and carried to index EPS by the 1990–2019 average gap between the two (−0.92 points a year). The middle 80% of paths runs from −9% to +22%; consensus is +15.1% on operating earnings, the model's on reported. In dollars the median path takes trailing EPS from $262.59 to $293.75 by Q2 2028. 80% of the paths fall below the consensus — the model's own share, not a probability: in the record its 80% band held 12 of 20. Drawn with the spread of that gap too (a standard deviation of 47.9 points over 120 quarters), the middle 80% runs from −44% to +96%; refitted each year-end, that wider band held 17 of 20 (Kupiec p 0.56) with a log score of −6.12. Still left out is the estimation itself — the likelihood has 6 local maxima, whose medians are +4.7%, +5.8%, +5.7%, +5.8%, +4.8%, +5.5%. The same model on trailing reported EPS gives +2.3%. Record, refitted at each year-end since Q4 2005 on the data to that date: the default fan's median missed the year's EPS growth by 49.8 log points (root mean square) against 49.4 for a fixed mean and 50.6 for the trailing-EPS model, and its 80% band held 12 of 20 outcomes (Kupiec p 0.04). Its log score, −12.37 against −7.45 for a fixed mean (higher is better), says the same as the coverage: the fan's density is too narrow and puts too little weight where outcomes landed; with the bridge drawn it scores −6.12. Read the 2027 figure as the model's, with that record.

  • Two horizons
    Over eight quarters the paths give 5.8% a year, Valuation's near-term growth, from a fan with no edge over a fixed mean.

    Calendar 2027 runs from Q4 2026 to Q4 2027: median EPS growth +4.7%. The eight quarters run from Q2 2026 to Q2 2028: median growth +11.9%, 5.77% a year compounded. They add Q3 2026, Q4 2026, Q1 2028, Q2 2028, where the median path grows at 6.8% a year against 4.8% inside 2027: one fan, read over two spans.

  • Cross-check
    A top-down regression puts calendar-2027 EPS growth at +14.1%, but out of sample it does worse than a forecast of zero growth.

    On Q1 1990 – Q4 2019, Q4 2008 – Q3 2010 excluded the regression explains 40% of the variance of EPS growth; each point of nominal GDP growth is worth 8.51 points of EPS growth (HAC SE 2.51, t 3.4); dollar −1.20 per % (t −1.8), WTI −0.040 (t −0.3), lagged margin gap 3.97 per pp (t 1.4). Since Q1 2022 its root-mean-square error is 21.8pp against 41.1pp for 'same growth as last year' and 17.5pp for zero growth (18 quarters to Q2 2026). Its calendar-2027 band across 4, 5 and 6% nominal growth is +5.6% to +22.6%, −1.0pp from the consensus. A model that does not beat 'no change' is a cross-check on the fan, not the page's expectation; its tables and fitted chart are in the Detail fold.

  • Trend
    The model puts nominal GDP's trend growth at 5.6% in Q2 2026, and the 4.0% assumption sits inside the band around it.

    The filtered long-run growth state (continuously compounded), 95% band 3.0 to 8.3. On nominal GDP: estimated freely, φ goes to its upper bound (0.99) while σ_η stays positive (0.090) — the state moves, and fixing φ at 0.95 imposes faster reversion to μ̄; it is not a state the data cannot find. It has been above the assumed rate since Q4 2020. The same model on real GDP and on prices puts real growth's trend at 2.0 (band 1.3–2.8) and the deflator's at 5.2 (band 4.4–6.1), 7.3 together — so most of the gap to 4.0 is prices. The LR test of the bound rule's φ on nominal GDP is 6.69 (p 0.010; free φ 0.990, at its bound): the data reject the fixed 0.95.

  • Growth
    Nominal GDP grew 6.3% in the four quarters to Q2 2026, against the 4.0% terminal growth the valuation assumes.

    In log points, the scale the model and the gaps use: nominal GDP +6.1, national-accounts profits after tax +16.7 and the index's trailing reported earnings per share +16.6 over the four quarters to Q2 2026. Since Q1 1960 the three have averaged 6.2, 7.2 and 6.5 log points a year.

  • Estimated rate
    Estimated rather than assumed, nominal GDP's long-run rate since 1990 is 5.0% with φ held, 4.8% with φ free.

    The same model with μ̄ freed: the profile likelihood over μ̄ on 146 quarters, φ at 0.95 as in the page's fit, the same jump dummies; standard error 1.89. The likelihood-ratio test of the assumed 4.0% gives 0.23 (p 0.63): the data do not reject it at 5%. The 95% likelihood interval is 2.50–8.75. With φ estimated too, on this sample it goes to its lower bound (0.80), the fastest reversion the model allows, and μ̂ to 4.8% (p 0.24 against 4.0). The 4.0% itself is the Committee's longer-run real GDP 2.0 + PCE inflation 2.0 (16 Sep 2026), the rate every other fit here reverts to.

  • Profits and EPS
    For profits and EPS the data cannot separate trend from cycle, so the model holds their trend at the assumed 4.0%.

    On profits after tax: estimated freely, σ_η goes to zero (0.000; φ 0.80) — the data cannot separate a slow trend from the cycle, so the trend is held at the assumed rate μ̄ and the spread of the paths is the cycle's and the jumps'; φ is then fixed at 0.95 by the rule. On index EPS: estimated freely, σ_η goes to zero (0.000; φ 0.80) — the data cannot separate a slow trend from the cycle, so the trend is held at the assumed rate μ̄ and the spread of the paths is the cycle's and the jumps'; φ is then fixed at 0.95 by the rule. The cycle absorbs their growth above the assumed rate: it averages +0.8 and +2.0 log points a year over the sample and stands at +5.4 and +28.8 in Q2 2026. That is a statement about the model on these series, not evidence that profits trend at 4.0: the gaps below measure what they have done.

  • Gaps
    Profits outgrew nominal GDP by 2.2 points a year in 1990–2019, but with a standard error of 1.5 the gap is not distinguishable from zero.

    Averages of four-quarter log growth differences over 120 quarters with Newey–West standard errors: profits less GDP +2.20 (1.54), index EPS less profits −0.92 (7.06, not distinguishable from zero either), EPS less GDP +1.28 (7.63) — the Earnings page's gaps, reproduced. A gap held for ever would put profits on a path to exceed the economy; over 1960–1989 profits less GDP averaged −0.6. In the year to Q2 2026: profits +10.6 over GDP, EPS −0.1 over profits.

Next: Discount rate — the rate these earnings are discounted at

Source: U.S. Securities and Exchange Commission (EDGAR: the members' filings; Form N-PORT); State Street Global Advisors; Yahoo Finance; BEA; Robert J. Shiller; Federal Reserve Board; EIA; Macro Review calculations · filings to Q2 2026cross.earnings_outlook-read

The index's trailing earnings, eight quarters on, from its own statements

Trailing four-quarter earnings per index unit on SPY's basis, history and the model's path with its band, beside the ten-year path

Exhibit 1 · S&P 500 trailing earnings and the model's projection, with bands

Dollars per index unit, four quarters to each quarter-end; SPY holdings, Shiller's series before; projection from Q2 2026 with 25th–75th and 10th–90th bands of draws
Range: simulated shocks
About
About this exhibit
Trailing four-quarter earnings per index unit in dollars, quarterly: history from the fund's holdings and Shiller's series before that, then the base projection for eight quarters with 25th–75th and 10th–90th bands of simulated draws
Trailing earnings per index unit, and the base's next eight quarters
Dollars per index unit, four quarters to each quarter's end: SPY's holdings back to 2020, Shiller's series before and beside it; then the base from Q2 2026, with the 25th–75th and 10th–90th of its draws.
The index's income statement, payout and balance sheet carried forward by their own equations: revenue from nominal GDP, the dollar and oil; a pretax margin that reverts toward its 2017–26 mean; dividends by Lintner's rule; buybacks, issuance and dilution to the share count. Model-conditional: what the statements imply if the equations hold, not a forecast of what will be filed.
Source: U.S. Securities and Exchange Commission (EDGAR: the members' filings; Form N-PORT); State Street Global Advisors; Yahoo Finance; BEA; Robert J. Shiller; Federal Reserve Board; EIA; Macro Review calculationscross.earnings_outlook-2

Exhibit 2 · Projected S&P 500 earnings per unit with growth, margin and payout

EPS, dollars per index unit, years from Q2 2026; growth, pretax margin and net payout, %; quarterly projection first, then a fade to long-run values
About
YearEPSGrowthMarginPayout
1 to Q2 2027$311.28+4.7%16.8%62.9%
2 to Q2 2028$310.17−1.9%15.5%67.5%
3 to Q2 2029$304.16−3.7%14.1%88.2%
4 to Q2 2030$317.96+2.2%13.7%79.7%
5 to Q2 2031$337.03+3.8%13.5%76.6%
6 to Q2 2032$358.99+4.4%13.4%76.0%
7 to Q2 2033$382.77+4.5%13.4%76.5%
8 to Q2 2034$407.82+4.4%13.4%77.4%
9 to Q2 2035$433.88+4.2%13.3%78.5%
10 to Q2 2036$460.71+4.0%13.3%78.7%

Growth is net income's; the payout is net of issuance and the clean-surplus residual. Terminal: 4.0% growth, 18.8% return on opening book, payout 78.7%.

About this exhibit
Table, year by year from the latest filed quarter: EPS per index unit, growth, pretax margin and net payout; the first years from the quarterly projection, the rest converging to long-run values
The path to the long run
Model years from Q2 2026; years 1–2 sum the explicit quarters, 3–10 fade to the terminal state.
Source: U.S. Securities and Exchange Commission (EDGAR: the members' filings; Form N-PORT); State Street Global Advisors; Yahoo Finance; BEA; Robert J. Shiller; Federal Reserve Board; EIA; Macro Review calculationscross.earnings_outlook-3

The statements behind the path

The income statement, the payout and share count, and the balance sheet with its return on equity, quarter by quarter then year by year

Exhibit 3 · Projected S&P 500 income statement, quarter by quarter then year by year

Dollars per index unit and %, by line of the statement; filed quarters Q3 2025 to Q2 2026, projected quarters to Q2 2028, then by year
PeriodRevenuePretax marginPretax incomeTaxesNet incomeEPSTrailing EPS
Q3 2025 (filed)$518.7–––$62.89$62.02–
Q4 2025 (filed)$536.4–––$64.61$63.98–
Q1 2026 (filed)$511.9–––$76.18$75.83–
Q2 2026 (filed)$546.8–––$90.74$90.74$292.58
Q3 2026$554.417.5%$97.1$18.8$77.45$77.78$308.33
Q4 2026$571.117.1%$97.6$18.9$77.85$78.49$322.84
Q1 2027$580.116.6%$96.3$18.6$76.78$77.71$324.72
Q2 2027$589.216.2%$95.4$18.4$76.08$77.30$311.28
Q3 2027$598.515.9%$94.9$18.3$75.68$77.20$310.70
Q4 2027$607.915.6%$94.7$18.3$75.53$77.36$309.57
Q1 2028$616.815.3%$94.6$18.3$75.45$77.59$309.45
Q2 2028$625.715.1%$94.7$18.3$75.55$78.02$310.17
Year 3$2,588.614.1%$364.9$70.6$291.07$304.16–
Year 4$2,730.013.7%$372.9$72.1$297.44$317.96–
Year 5$2,872.413.5%$387.0$74.8$308.72$337.03–
Year 6$3,015.313.4%$403.9$78.1$322.21$358.99–
Year 7$3,157.913.4%$422.0$81.6$336.67$382.77–
Year 8$3,299.613.4%$440.6$85.2$351.46$407.82–
Year 9$3,439.613.3%$459.1$88.8$366.27$433.88–
Year 10$3,577.213.3%$477.5$92.3$380.92$460.71–
About this exhibit
Table of the projected income statement per index unit: revenue, pretax margin, pretax income, taxes, net income, EPS and trailing EPS; the filed quarters, the projected quarters, then the later years
The income statement
Per index unit at the start; quarters as amounts for the quarter, years 3–10 for the year. The filed quarters are shown for the start.
Source: U.S. Securities and Exchange Commission (EDGAR: the members' filings; Form N-PORT); State Street Global Advisors; Yahoo Finance; BEA; Robert J. Shiller; Federal Reserve Board; EIA; Macro Review calculationscross.earnings_outlook-4

Exhibit 4 · Projected S&P 500 dividends and buybacks, with payout and share count

Dollars per index unit, % of earnings and share count (1 at Q2 2026); filed quarters to Q2 2026, projected to Q2 2028, then by year
PeriodDividendsBuybacksIssuancePayout, netPayout, grossShare countDividend per unit
Q3 2025 (filed)$20.14$26.25$1.82––1.0140$19.86
Q4 2025 (filed)$19.39$27.07$1.66––1.0098$19.20
Q1 2026 (filed)$21.25$33.98$1.49––1.0046$21.16
Q2 2026 (filed)$19.94$28.89$1.82––1.0000$19.94
Q3 2026$20.64$30.75$2.8661.4%66.3%0.9958$20.73
Q4 2026$21.33$31.10$3.0062.2%67.3%0.9918$21.50
Q1 2027$21.98$30.87$3.1363.4%68.8%0.9880$22.25
Q2 2027$22.61$30.78$3.1464.6%70.2%0.9843$22.97
Q3 2027$23.21$30.81$2.9966.0%71.4%0.9803$23.68
Q4 2027$23.80$30.95$2.9867.1%72.5%0.9764$24.37
Q1 2028$24.37$31.10$2.9568.1%73.5%0.9724$25.06
Q2 2028$24.92$31.34$2.9469.0%74.5%0.9684$25.74
Year 3$93.69$179.40$11.7488.2%93.8%0.9456$97.90
Year 4$96.70$156.40$11.3179.7%85.1%0.9254$103.38
Year 5$101.37$151.60$11.5676.6%81.9%0.9066$110.67
Year 6$106.85$155.11$12.0076.0%81.3%0.8885$119.05
Year 7$112.74$162.58$12.5276.5%81.8%0.8706$128.18
Year 8$118.83$171.97$13.0877.4%82.7%0.8530$137.89
Year 9$125.03$182.15$13.6678.5%83.9%0.8354$148.11
Year 10$131.27$188.99$14.2378.7%84.1%0.8182$158.77

The share count is one at the start quarter's end. The net payout leaves out issuance and the clean-surplus residual; the gross payout (dividends and buybacks over earnings) adds them back.

About this exhibit
Table per index unit: dividends, buybacks, issuance, net and gross payout, share count and dividend per unit; filed quarters, projected quarters, then the later years; net payout excludes issuance
Payout and the share count
Dividends by Lintner's rule, buybacks a fading share of earnings, issuance a share of market value; the share count falls by net buybacks over market value.
Source: U.S. Securities and Exchange Commission (EDGAR: the members' filings; Form N-PORT); State Street Global Advisors; Yahoo Finance; BEA; Robert J. Shiller; Federal Reserve Board; EIA; Macro Review calculationscross.earnings_outlook-5

Exhibit 5 · Projected S&P 500 balance sheet and return on equity

Dollars per index unit and %: book value, assets, debt and return on equity; filed quarters to Q2 2026, projected to Q2 2028, then by year
PeriodBook valueResidualBook per unitROE, average bookROE, opening bookAssetsDebt
Q3 2025 (filed)$1,247.9–$1,230.6–––$1,017
Q4 2025 (filed)$1,270.1–$1,257.8–––$1,035
Q1 2026 (filed)$1,297.1–$1,291.2–––$1,057
Q2 2026 (filed)$1,331.3–$1,331.3––$6,160$1,085
Q3 2026$1,361.2$1.01$1,366.923.7%–$6,264$1,109
Q4 2026$1,390.7$1.03$1,402.124.2%–$6,365$1,133
Q1 2027$1,418.8$1.05$1,436.023.8%–$6,564$1,156
Q2 2027$1,445.7$1.07$1,468.822.2%–$6,688$1,178
Q3 2027$1,471.5$1.09$1,501.021.6%–$6,817$1,199
Q4 2027$1,496.3$1.11$1,532.521.1%–$6,924$1,219
Q1 2028$1,520.4$1.13$1,563.520.6%–$7,031$1,239
Q2 2028$1,543.8$1.15$1,594.220.2%–$7,137$1,258
Year 3$1,578.2$4.67$1,669.018.6%18.9%$7,544$1,286
Year 4$1,638.6$4.77$1,770.818.5%18.8%$7,956$1,335
Year 5$1,710.9$4.96$1,887.118.4%18.8%$8,371$1,394
Year 6$1,788.3$5.18$2,012.718.4%18.8%$8,787$1,457
Year 7$1,867.6$5.41$2,145.018.4%18.8%$9,203$1,522
Year 8$1,947.0$5.65$2,282.618.4%18.8%$9,616$1,586
Year 9$2,025.6$5.89$2,424.818.4%18.8%$10,024$1,650
Year 10$2,106.6$6.13$2,574.618.4%18.8%$10,425$1,716
About this exhibit
Table per index unit: book value, clean-surplus residual, book per unit, return on equity on average and on opening book, assets and debt; filed quarters, projected quarters, then the later years
The balance sheet and the return on equity
Book value rolls forward by clean surplus, so the return on equity is an output; assets at the start's turnover, debt at its ratio to book.
Source: U.S. Securities and Exchange Commission (EDGAR: the members' filings; Form N-PORT); State Street Global Advisors; Yahoo Finance; BEA; Robert J. Shiller; Federal Reserve Board; EIA; Macro Review calculationscross.earnings_outlook-6

The parameters, and what the path assumes about the ten largest companies

Every parameter with its error, window and source; the same equations on the ten largest companies and on the rest

Exhibit 6 · Earnings model parameters with their standard errors, sample windows and sources

One row per parameter: value (ratio or %), standard error, estimation window and data source; the longest window Q2 1947 – Q2 2026
Range: estimation error
SymbolWhatValueStd. errorWindowSource
β_gRevenue's elasticity to nominal GDP
estimated 1.256 (se 0.048), shrunk toward 1 with a prior sd of 0.10
1.2080.043Q2 1947 – Q2 2026NIPA Table 1.14 line 17 on Table 1.1.5 line 1, quarterly log changes
δRevenue drift, a quarter (the members' share drift)
index basis alone 0.07% a quarter
0.001730.00125Q1 2017 – Q2 2026the index basis's revenue, four-quarter log changes, Q2 2021 on; company totals (today's members) Q1 2017 – Q1 2021, stated; aggregate = per unit + the measured share-count effect
β_$Revenue's elasticity to the broad dollar
index basis alone 0.059
0.0200.077Q1 2017 – Q2 2026the index basis's revenue, four-quarter log changes, Q2 2021 on; company totals (today's members) Q1 2017 – Q1 2021, stated; aggregate = per unit + the measured share-count effect
β_oRevenue's elasticity to WTI
index basis alone 0.071
0.0590.016Q1 2017 – Q2 2026the index basis's revenue, four-quarter log changes, Q2 2021 on; company totals (today's members) Q1 2017 – Q1 2021, stated; aggregate = per unit + the measured share-count effect
μPretax margin's mean
anchors: 11.8% (2017–19), 14.3% (since 2021)
13.35%–Q1 2017 – Q2 2026, 4 pandemic quarters outone-quarter pretax income ÷ revenue: the index basis from Q3 2019, company totals before (spliced by the median log gap)
φMargin's persistence, a quarter
with the pandemic's quarters in: 0.676
0.8430.117Q1 2017 – Q2 2026, 4 pandemic quarters outone-quarter pretax income ÷ revenue: the index basis from Q3 2019, company totals before (spliced by the median log gap); a measurement equation for the filings' one-quarter noise
γMargin's response to revenue growth (operating leverage)
with the pandemic's quarters in: 2.41
0.5990.644Q1 2017 – Q2 2026, 4 pandemic quarters outone-quarter pretax income ÷ revenue: the index basis from Q3 2019, company totals before (spliced by the median log gap)
σ_εThe filings' one-quarter margin noise
enters the band: an outcome is a filed figure
0.106–Q1 2017 – Q2 2026, 4 pandemic quarters outone-quarter pretax income ÷ revenue: the index basis from Q3 2019, company totals before (spliced by the median log gap)
σ_SRevenue shock, a quarter
Student-t, 5.9 degrees of freedom
0.00854–Q2 1947 – Q2 2026the residuals of β_g's regression on NIPA (gross value added on GDP)
σ_mMargin shock, a quarter
Student-t, 4.5 degrees of freedom; φ there 0.957
0.051–Q2 1947 – Q2 2026Table 1.14: nonfinancial profits ÷ gross value added, AR(1) with operating leverage
τTax rate
the statutory 21% −1.7pp, held (no change in law)
19.34%–Q3 2025 – Q2 2026the index basis: income taxes ÷ pretax income, four quarters
oShare of after-tax income reaching shareholders98.89%–Q3 2025 – Q2 2026the index basis: net income ÷ (pretax income − taxes), four quarters
cDividends' adjustment speed, a quarter (Lintner)
the index basis Q2 2020 – Q2 2026: 0.025 (se 0.015)
0.0260.011Q1 1990 – Q2 2026Shiller's dividends and earnings per share at quarter-end months
d*Dividends' target payout (Lintner)
the index basis: 52.3% (se 10.2pp)
61.12%10.70ppQ1 1990 – Q2 2026Shiller's dividends and earnings per share at quarter-end months
b₀Buybacks ÷ earnings, the trailing four quarters39.44%–Q2 2026the index basis
sDividends' share of the payout (the terminal split)40.99%–Q2 2026the index basis, four quarters
ιIssuance ÷ market value, a year0.15%0.02ppQ2 2020 – Q2 2026the index basis: issuance proceeds ÷ the close, four quarters
νNon-cash dilution, a year
net buyback yield 1.73%, share-count effect 1.91%
-0.18%–Q2 2020 – Q2 2026the measured share-count effect (company totals) less the net buyback yield (the index basis)
x̄Clean-surplus residual ÷ opening book, a year
on company totals (today's members, through their acquisitions) 3.8% a year
0.30%0.78ppQ4 2019 – Q2 2026the index basis: Δbook − (earnings − dividends − buybacks + issuance), the count by the share identity
θAsset turnover (revenue ÷ assets)0.343–Q2 2026the index basis, held
λDebt ÷ book value0.815–Q2 2026the index basis, held
—Cash ÷ assets6.50%–Q2 2026the index basis, held
g_LLong-run nominal growth (the terminal)4.00%–SEP of 16 Sep 2026the SEP's longer-run real GDP growth + PCE inflation
ROE*Long-run return on equity, opening book (the terminal)
sensitivities: 15.9% (2017–19, company totals) and today's 24.2%
18.81%1.32ppQ3 2020 – Q2 2026the index basis: four quarters' earnings ÷ the book a year before
k*Terminal payout net of issuance and the residual: 1 − g_L ÷ ROE*78.73%–—the identity
About this exhibit
Table of the earnings model's parameters: symbol, meaning, value, standard error, estimation window and source; revenue elasticities to GDP, the dollar and oil; the margin's mean, persistence and operating leverage; shocks; tax; and payout
The parameters, with their errors, windows and sources
Each estimated on the window named; the error scales from the long record, never from the short sample's residuals.
Source: U.S. Securities and Exchange Commission (EDGAR: the members' filings; Form N-PORT); State Street Global Advisors; Yahoo Finance; BEA; Robert J. Shiller; Federal Reserve Board; EIA; Macro Review calculationscross.earnings_outlook-7

Exhibit 7 · The largest companies against the rest: implied earnings, margin and share

Dollars per index unit and %, 10 years from Q2 2026: the rest on its own equations, the ten largest what the index projection leaves
About
YearIndex net incomeThe rest, own runThe ten, impliedThe ten's margin, impliedThe ten's share, impliedThe buckets' own sumAggregation effect
1$308.2$198.4$109.736.8%35.6%$319.9−11.8
2$302.2$209.2$93.029.4%30.8%$321.1−18.9
3$291.1$211.7$79.323.8%27.3%$314.9−23.8
4$297.4$219.5$78.022.2%26.2%$323.6−26.2
5$308.7$229.2$79.521.5%25.8%$336.7−28.0
6$322.2$239.8$82.421.3%25.6%$351.8−29.6
7$336.7$250.8$85.821.2%25.5%$367.7−31.0
8$351.5$262.0$89.521.1%25.5%$383.9−32.5
9$366.3$273.0$93.221.1%25.5%$400.1−33.9
10$380.9$284.0$97.021.1%25.5%$416.1−35.2

Own levels — the ten: margin mean 28.4%, Lintner target 27%, long-run return 44.0%; the rest: 11.8%, 61%, 15.8%. The effect is the index path less the buckets' own sum, in dollars per index unit.

About this exhibit
Table year by year, in dollars per index unit and %: index net income; the rest's own projection; the largest companies' implied earnings, margin and share of index earnings; the two groups' own sum; and the aggregation effect
What the index path assumes about the ten largest companies
The ten by the fund's value on 30 Jun 2026 (37.9% of it): Apple Inc, Nvidia Corp, Microsoft Corp, Alphabet Inc, Amazon.com Inc, Broadcom Inc, Meta Platforms Inc, Eli Lilly & Co, Tesla Inc, Micron Technology Inc. Their trailing share of index earnings is 39.4%.
Source: U.S. Securities and Exchange Commission (EDGAR: the members' filings; Form N-PORT); State Street Global Advisors; Yahoo Finance; BEA; Robert J. Shiller; Federal Reserve Board; EIA; Macro Review calculationscross.earnings_outlook-8

The record, the other readings and the sensitivities

How the model did at each quarter since 2021 on the data then filed; the statistical and market-implied legs and the dated outside readings; the named alternative runs

Exhibit 8 · Earnings model's forecast errors against simple benchmarks, by horizon

Out-of-sample errors since Q2 2021, log points (≈ %), at 1, 4 and 8 quarters ahead: the model, last year's growth and Shiller's trend; with band coverage
Range: out-of-sample recordAbout
Quarters aheadOriginsModel RMSEBiasLast year's growthShiller's trend80% band heldBefore wideningKupiec pPIT, fifths
1202.8−0.56.45.190%85%0.232 · 3 · 6 · 7 · 2
41710.4+2.330.911.288%59%0.373 · 7 · 1 · 4 · 2
81315.9+7.568.313.1100%38%0.024 · 3 · 4 · 1 · 1

Thin: 13 origins at eight quarters, overlapping. The switching rule — the structural leg replaces the statistical one as the near-term growth in the Valuation page's bridge when its eight-quarter error is lower over the newest 12 published origins, consecutive quarters — is not met: 0 published paths have their outcome, the first due Q2 2028.

About this exhibit
Table of the out-of-sample record at each horizon: origins, the model's root mean square error and bias, the errors of last year's growth and of Shiller's trend, band coverage before and after widening, Kupiec p and PIT
The record since Q2 2021
At each quarter: the model started from the filings as first reported and run on the macro path as it came; errors in log points (forecast over outcome, ×100).
Source: U.S. Securities and Exchange Commission (EDGAR: the members' filings; Form N-PORT); State Street Global Advisors; Yahoo Finance; BEA; Robert J. Shiller; Federal Reserve Board; EIA; Macro Review calculationscross.earnings_outlook-9

Exhibit 9 · S&P 500 earnings growth on the site's estimates and outside forecasts

EPS growth, %, for 2026, 2027 and eight quarters: this projection, the site's other estimates, FactSet consensus (2 Oct 2026) and the Livingston Survey (Jun 2026)
Leg or reading20262027Eight quartersBasis and date
The base (this page)+34.3%−4.1%+6.3%GAAP, the index basis; calendar years add the filed quarters to the base's
The statistical leg (Earnings)–+4.7%+11.9%GAAP, Shiller's series; from Q2 2026
The market-implied leg (Valuation)5.7% a year, long runthe growth today's price implies; Sep 2026
FactSet, bottom-up consensus+32.4%––operating, FactSet's aggregation; 2 Oct 2026 (past its age) · forward P/E 19.0
Livingston Survey, after-tax profits–+4.2%–after-tax profits in the national accounts, the median forecaster, +5.4% over twelve months; survey of Jun 2026
About this exhibit
Table of EPS growth for this year, next year and eight quarters, each on its own basis, from this page's projection, the Earnings page's statistical estimate, the Valuation page's market-implied rate, FactSet bottom-up consensus and the Livingston Survey
Three legs, and the readings beside them
Growth in earnings per share; each leg on its own basis, said in its row. The outside readings are dated quotes, never inputs.
Source: U.S. Securities and Exchange Commission (EDGAR: the members' filings; Form N-PORT); State Street Global Advisors; Yahoo Finance; BEA; Robert J. Shiller; Federal Reserve Board; EIA; Macro Review calculationscross.earnings_outlook-10

Exhibit 10 · Projected earnings with margin and payout when one assumption changes

EPS in dollars per index unit at years 1, 2 and 10 from Q2 2026; margin, return on equity and payout, %; first row: base case
Range: assumed values
RunEPS, year 1EPS, year 2Margin, year 2EPS, year 10ROE, year 10 (average book)Payout, year 10Per-share growth after
The base$311.28$310.1715.5%$460.7118.4%78.7%6.18%
μ at the 2017–19 mean$298.68$285.6614.3%$414.9518.4%78.7%6.18%
μ at the mean since 2021$318.40$324.4216.2%$488.2018.4%78.7%6.18%
ROE* at the 2017–19 level$311.27$310.1315.5%$440.3615.5%74.8%6.08%
ROE* at today's$311.29$310.2215.5%$486.7223.7%83.5%6.29%
The market value at a constant price$311.41$310.5815.5%$467.1618.4%78.7%6.75%
A five-year fade$311.28$310.1715.5%$442.5518.4%78.7%6.18%
The margin fitted with the pandemic's quarters in$313.60$295.3014.7%$438.9818.4%78.7%6.18%
The start at the filed margin, not the filtered$346.05$327.2716.3%$462.7418.4%78.7%6.18%
About this exhibit
Table of alternative runs, each changing one assumption (margin mean, long-run return on equity, market value, fade length, pandemic quarters or starting margin): EPS at three years, the year-two margin, the terminal return on equity and payout, and per-share growth
The named alternative runs
Each changes one assumption and holds the rest; the first row is the base.
Source: U.S. Securities and Exchange Commission (EDGAR: the members' filings; Form N-PORT); State Street Global Advisors; Yahoo Finance; BEA; Robert J. Shiller; Federal Reserve Board; EIA; Macro Review calculationscross.earnings_outlook-11

The record by origin, the margin's long record and the run log

Each origin's forecast against its outcome; the margin equation on the national accounts since 1970; every path as published

Exhibit 11 · Four-quarter-ahead earnings forecasts and bands against actual outcomes

Trailing earnings, dollars per index unit, quarterly origins since Q2 2021; black dot inside the band drawn four quarters earlier, red outside
Range: out-of-sample record
About this exhibit
Chart of trailing earnings per index unit since 2021: at each quarterly origin, the band drawn four quarters earlier against the outcome; a black dot when the outcome is inside the band, red when outside
Four quarters ahead, at each origin, against the outcome
Trailing earnings per index unit; dot black inside the band drawn four quarters earlier, red outside.
Source: U.S. Securities and Exchange Commission (EDGAR: the members' filings; Form N-PORT); State Street Global Advisors; Yahoo Finance; BEA; Robert J. Shiller; Federal Reserve Board; EIA; Macro Review calculationscross.earnings_outlook-12

Exhibit 12 · Margin model's forecast errors and band coverage on national accounts data

Nonfinancial corporate profits over gross value added (BEA), refitted each quarter since Jan 1970; 1, 4 and 8 quarters ahead; errors in log points (≈ %)
Range: out-of-sample record
Quarters aheadOrigins80% band heldKupiec pRMSENo changeWidening needed
122571%<0.015.86.3×1.23
422256%<0.0113.013.1×1.74
821843%<0.0120.819.3×2.26

Errors in log points of the margin. The page's band is widened by these factors, interpolated across the quarters between them; the index's own record above is the check on that choice, not its source.

About this exhibit
Table of the margin equation's out-of-sample record on national accounts data (nonfinancial profits over gross value added) at each horizon: origins, band coverage, Kupiec p, errors against a no-change forecast and the widening the band needs
The margin equation on the national accounts since 1970
Nonfinancial corporate profits over gross value added (Table 1.14), the equation refitted at each quarter on the data to then and run on the value added that came.
Source: U.S. Securities and Exchange Commission (EDGAR: the members' filings; Form N-PORT); State Street Global Advisors; Yahoo Finance; BEA; Robert J. Shiller; Federal Reserve Board; EIA; Macro Review calculationscross.earnings_outlook-13

Exhibit 13 · Log of every published earnings projection, never revised

Trailing earnings, dollars per index unit, 1, 4 and 8 quarters ahead, with the statistical estimate; one row per publication since 12 Oct 2026
RowLoggedFrom1 quarter4 quarters8 quartersStatistical leg, 8 quarters
112 Oct 2026Q2 2026$309.42$314.09$311.80$327.55
226 Oct 2026Q2 2026$309.29$314.02$311.93$327.49
331 Oct 2026Q2 2026$308.68$312.24$310.32$327.33
About this exhibit
Table logging each published projection: date logged, origin quarter, trailing earnings at each horizon and the statistical estimate at eight quarters
Every path as published
One row per published path, never rewritten; the switching rule reads these, not the record above.
Source: U.S. Securities and Exchange Commission (EDGAR: the members' filings; Form N-PORT); State Street Global Advisors; Yahoo Finance; BEA; Robert J. Shiller; Federal Reserve Board; EIA; Macro Review calculationscross.earnings_outlook-14

What does the model expect index earnings to do?

The model's simulated paths for index earnings run forward from the last reported quarter, in dollars and as growth, their quantiles, and the same paths read over the calendar year and over eight quarters — , not a forecast

Exhibit 14 · S&P 500 EPS: reported history and simulated paths, against consensus

$ a share, quarterly: Shiller reported EPS, then 20,000 simulated paths to Q2 2028; dots: FactSet calendar-year consensus, operating earnings (a higher basis); calendar 2027 shaded
About this exhibit
Chart: Shiller reported EPS, $ a share, history then the model's simulated paths quarterly, with FactSet calendar-year consensus dots (operating earnings) and the calendar year shaded; switchable to growth.
The EPS fan in dollars: reported earnings and the model's paths
Reported EPS (Shiller) and 20,000 simulated paths, dollars a share, quarterly to Q2 2028; the consensus dots are FactSet's operating earnings, a different basis from the fan's reported EPS.
, not a forecast: the of national-accounts profits, run forward from Q2 2026 and carried to EPS by the 1990–2019 average gap between EPS and profits growth (−0.92 points a year). The bands are the model's own shocks carried forward past misses; they leave out the spread of that gap (drawn in, the calendar 2027 band widens to −44% to +96%: the robustness table's second row) and the choice among the likelihood's local maxima. The dots are FactSet's calendar-year consensus: earnings, which run above the reported earnings the fan is drawn in, so read the distance between them as partly a difference of basis. Shaded: Calendar 2027, whose growth is the model's +4.7% (80% of paths below the +15.1% consensus).
Source: Robert J. Shiller; U.S. Bureau of Economic Analysis; FactSet Earnings Insight; Macro Review calculations · data to Q2 2026cross.earnings-7

Exhibit 15 · Percentiles of simulated profits and EPS growth, four and eight quarters ahead

%, percentiles of the simulated paths: profits growth over four and eight quarters, and calendar 2027 carried to EPS (last row).
Measure5th10th25thMedian75th90th95th
Quarterly profits growth, annualised — 4 quarters aheadlog points a year−26.4−19.3−7.7+5.1+18.5+30.1+37.8
Quarterly profits growth, annualised — 8 quarters aheadlog points a year−25.6−18.7−6.7+6.2+19.2+31.2+38.9
Cumulative profits growth over 4 quarters%−11.3−7.3−0.6+7.5+16.1+25.0+31.1
Cumulative profits growth over 8 quarters%−13.4−7.8+2.2+13.9+27.2+41.5+51.0
Profits growth over calendar 2027%−12.5−8.5−1.9+5.7+14.2+22.9+29.0
EPS growth over calendar 2027, bridged (the consensus comparison)%−13.3−9.3−2.8+4.7+13.1+21.8+27.8
About this exhibit
Table: percentiles of the simulated paths for profits growth over four and eight quarters and over the calendar year, and the calendar year's EPS growth.
The fan in numbers: quantiles of profits and EPS growth
Percentiles of the simulated paths: profits growth over four and eight quarters, and calendar 2027 carried to EPS (last row).
Consensus for calendar 2027 is +15.1%; 79.6% of the paths carried to EPS land below it. The same model on trailing EPS is the robustness case, in the Detail fold. past misses
Source: U.S. Bureau of Economic Analysis; Robert J. Shiller; Federal Reserve (Summary of Economic Projections); Macro Review calculations · data to Q2 2026cross.earnings-9

Exhibit 16 · Median EPS growth over the calendar year and over eight quarters

Median of the same simulated paths over both spans: log points, then EPS growth and its annual rate, %; Q2 2026 to Q2 2028.
SpanProfits, log pointsBridge, log pointsEPS growth, %A year, %
Calendar 2027Q4 2026 → Q4 2027 · this page: the comparison with the calendar-year consensus+5.54−0.92+4.74.72
Eight quarters from Q2 2026Q2 2026 → Q2 2028 · the Valuation page: near-term growth in the two-stage model+13.06−1.84+11.95.77
About this exhibit
Table: the paths' median over two spans, the calendar year and the eight quarters: profits in log points, the EPS bridge, EPS growth and its annual rate.
One fan, two horizons: the calendar year and the eight quarters
Median of the same simulated paths over two spans: log points, then EPS growth and its annual rate; Q2 2026 to Q2 2028.
The calendar year is compared with consensus on this page; the eight quarters are annualised into the near-term growth of the Valuation page's two-stage model. The eight quarters add Q3 2026, Q4 2026, Q1 2028, Q2 2028, where the median path grows at 6.8% a year against 4.8% inside the calendar year. past misses
Source: U.S. Bureau of Economic Analysis; Robert J. Shiller; Macro Review calculations · data to Q2 2026cross.earnings-25

The fan's robustness and record, and a regression's band

The fan on profits against the same model on trailing EPS, the calendar-year figure when one thing changes, each year-end refit scored against what happened, and a top-down regression's calendar-year band with its record

Exhibit 17 · Calendar-year EPS growth from the profits model and the trailing-EPS model

Calendar 2027 EPS growth, %: percentiles of the paths and the share below the +15.1% consensus, fit by fit.
About
Fit10thMedian90thBelow consensus
NIPA profits after tax, bridged to EPS (the default fan)profits growth shifted by −0.92 log points, the 1990–2019 mean EPS-minus-profits wedge−9.3%+4.7%+21.8%80%
… with the bridge's spread drawneach path's gap a normal draw, standard deviation 47.9 log points−44.4%+4.3%+96.2%58%
… before the bridge (profits growth)one quarter's flow, non-overlapping−8.5%+5.7%+22.9%77%
Trailing 12-month reported EPS (robustness case)the quarterly change of a four-quarter moving sum−29.8%+2.3%+49.4%66%
Switching-variance preset (trailing EPS, robustness)the fitted Markov chain simulated directly−4.2%+7.9%+21.0%77%

Jumps on the non-overlapping series: Q4 2008, Q1 2009, Q3 2020 (p 1.10% a quarter), against 4 on the trailing one. The bridge is a level shift in logs, so it moves every percentile of the fan by the same amount (a check); its standard deviation over the 120 quarters is 47.9 log points, which is the dispersion the bridge does not carry. The NIPA fit is the best of 29 starting points: its likelihood has 6 local maxima, and the default fan's median at each is +4.7%, +5.8%, +5.7%, +5.8%, +4.8%, +5.5% — estimation uncertainty the bands above do not carry.

About this exhibit
Table: calendar-year EPS growth (10th, 50th, 90th percentiles) and the share below consensus for the profits model, its variants and the trailing-EPS model.
The fan on profits against the same model on trailing EPS
calendar 2027 EPS growth, 10th / 50th / 90th of the paths, and the share below the +15.1% consensus, fit by fit.
calendar 2027 growth, 10th / 50th / 90th of the simulated paths, and the share of paths below the quoted consensus of +15.1%. NIPA after-tax profits are the quarter's own flow (266 quarters from Q1 1960, anchor 6.20 log points a year over the fan's horizon, cycle period 10.8 quarters); the trailing series is a four-quarter moving sum — consecutive observations share three of four quarters — so an AR(2) fitted to it recovers part of the filter. past misses the model's own shocks, carried forward; the robustness rows other models
Source: U.S. Bureau of Economic Analysis; Robert J. Shiller; Federal Reserve (Summary of Economic Projections); Macro Review calculations · data to Q2 2026cross.earnings-8

Exhibit 18 · Calendar-year EPS growth when one model choice changes

Calendar 2027 EPS growth, %, from the profits model, then the trailing-EPS model and its variants, each re-run with one change.
Setting10thMedian90thP(< consensus)
Default: the non-overlapping NIPA fit, bridged to EPS−9.3%+4.7%+21.8%79.6%
Robustness case: the trailing-series fan (jump-adjusted, φ fixed at 0.95)−29.8%+2.3%+49.4%66.1%
Trailing series, assumed long-run rate −0.5pp (μ̄ 4.78 log points a year)−30.2%+1.7%+48.7%66.8%
Trailing series, assumed long-run rate +0.5pp (μ̄ 5.78 log points a year)−29.4%+2.8%+50.2%65.4%
Trailing series, φ fixed at 0.90−29.8%+2.2%+49.4%66.2%
Trailing series, φ fixed at 0.95−29.8%+2.3%+49.4%66.1%
Trailing series, φ estimated freely (it goes to 0.80, the bound)−29.7%+2.2%+49.2%66.2%
Trailing series, no jumps (p = 0)−26.8%+2.9%+45.4%66.4%
Trailing series, jumps resampled symmetrically (|size| with a random sign)−29.2%+3.0%+49.9%65.3%
Trailing series, Markov-switching variance (σ 3.9 / 38.1, P(high) 1%)−27.5%+3.1%+46.1%65.6%
About this exhibit
Table: calendar-year EPS growth (10th, median, 90th) and share below consensus for the default model and its variants, each with one change.
Robustness: the calendar-year figure when one thing changes
calendar 2027 EPS growth from the fan, then the trailing-EPS model and its variants, each re-run with one change.
other models — the same question answered by other fits.
Source: U.S. Bureau of Economic Analysis; Robert J. Shiller; Federal Reserve (Summary of Economic Projections); Macro Review calculations · data to Q2 2026cross.earnings-10

Exhibit 19 · How the earnings models forecast, against a random walk and fixed mean

20 scored four-quarter EPS-growth forecasts from year-ends Q4 2005–Q4 2024: RMSE in log points and % growth, log score, sample
About
ForecasterRMSE, log pointsRMSE, % growthLog scoreSample
The default fan: profits bridged to EPS49.864.2pp−12.3720 year-end origins, Q4 2005 – Q4 2024; 80% band held 12 of 20 (Kupiec p 0.04); mean error +1.6
The trailing-EPS model (robustness case)50.669.3pp−6.8220 year-end origins, Q4 2005 – Q4 2024
Random walk with drift179.0not comparable−10.63same origins; its log forecast reaches 568 log points once
Fixed mean49.463.4pp−7.45same origins
The macro regression above (M3)–21.8pp–18 quarters, Q1 2022 – Q2 2026 (its own window)
Both on the quarters they share–49.4pp / 24.5pp–process / macro, 6 fourth quarters: Q4 2020, Q4 2021, Q4 2022, Q4 2023, Q4 2024, Q4 2025
80% fan coverage70% (14 of 20 inside; nominal 80%)6 outside

The default fan's record: at each year-end from Q4 2005 the fan's whole construction is repeated on the data to that date only — the profits-minus-GDP wedge and the EPS-minus-profits bridge averaged from 1990 to the origin (to 2019 at the latest), the anchor from them, the bound rule, the jump dummies and the fixed start grid (609 fits converged, 0 failed) — and four quarters of profits growth simulated (20,000 paths) and carried to EPS by that origin's bridge. It is scored on the same origins and target as the trailing model below: RMSE 49.8 log points against 49.4 for a fixed mean, mean error +1.6, 80% band 12 of 20 (Kupiec p 0.04). Its band carries the simulation's spread only — not the bridge's own spread or the estimation's — and the record shows what that costs. Recursive out of sample: the whole two-step estimation is repeated on the data to each year-end from 2005 to 2025 (21 refits) and the next four quarters' cumulative growth simulated (20,000 paths). 20 of those have all four quarters observed and are scored; 1 is still pending. Benchmarks on the same origins: a random walk with drift on y (ŷ_{T+h} = y_T + hδ̂, δ̂ the mean change of y in the training sample, predictive variance 30 δ̂-variances for the four-quarter sum) and a fixed mean (the training sample's mean four-quarter sum with its own sample variance). Point forecasts are each method's median, so the percentage column is the log column transformed; RMSE is reported in log points (the headline) and on simple four-quarter percentage growth for comparability with the macro regression above, and the log score is the Gaussian predictive density at the realisation on the log scale (higher is better). The random walk's four-quarter log forecast reaches 568 log points at one origin (it carries a single quarter of 2009 forward four times), so its percentage-scale error is arithmetic, not a comparable number: read that benchmark in log points. The macro regression's out-of-sample RMSE (21.8pp, Q1 2022 – Q2 2026) is a different window and a different object — quarterly four-quarter growth from realised macro inputs — so the table also prints both models on the 6 year-end quarters they share. The assumed long-run rate is today's SEP median at every origin, not the SEP of the day: the backtest tests the process, not real-time anchor selection.

About this exhibit
Table: forecast accuracy (RMSE, log score) of the two earnings models refitted at each year-end, against a random walk and a fixed mean.
The fans' records: one refit per year-end
20 scored four-quarter forecasts of EPS growth from Q4 2005: the default fan and the trailing-EPS model, each refitted on the data to the origin, against a random walk and a fixed mean.
21 refits from Q4 2005 to Q4 2025; 20 have all four quarters observed and are scored. Point forecasts are each method's median; the log score is the Gaussian predictive density at the realisation (higher is better). out-of-sample record
Source: U.S. Bureau of Economic Analysis; Robert J. Shiller; Federal Reserve (Summary of Economic Projections); Macro Review calculations · data to Q2 2026cross.earnings-11

Exhibit 20 · Calendar-year EPS growth under each nominal growth rate, against consensus

4-quarter EPS growth at Q4 2027 under nominal GDP growth of 4, 5 and 6%, against consensus and the regression's record since Q1 2022.
About
Calendar 2027, ex-2008–104% nominal GDP5% (mid)6%ConsensusMid − consensusOut of sample since Q1 2022
EPS growth+5.6%+14.1%+22.6%+15.1%−1.0pp (inside ± 21.8pp)RMSE 21.8pp vs 17.5pp for zero growth
About this exhibit
One-row table: calendar-year EPS growth under each nominal GDP growth scenario, consensus, the gap and the out-of-sample record.
A top-down regression's calendar-2027 band
EPS growth under 4, 5 and 6% nominal GDP growth, against consensus and the regression's record since Q1 2022.
The 4–6% columns are scenarios named paths, no odds; the out-of-sample error is a backtest out-of-sample record. Conditional arithmetic on stated nominal-growth paths, not a forecast; the model's expectation is the fan. The regression's coefficients, its out-of-sample quarters and its fitted chart are in the Earnings page's Detail fold; its eight-quarter paths are on the Scenarios page.
Source: U.S. Bureau of Economic Analysis; Robert J. Shiller; Federal Reserve Board (H.10); EIA; FactSet Earnings Insight; Macro Review calculations · data to Q2 2026cross.earnings-24

The earnings path the assumed rate implies over the fade

Trailing EPS grown along the two-stage model's fade to , at the assumed rate and a point either side — , not a forecast

Exhibit 21 · Trailing EPS as growth slows to each terminal rate

$ a share, year by year from Q3 2026: growth fading over six years to five terminal rates; bold, the assumed rate
About
Year toGrowth at 4.0EPS, terminal 3.0%EPS, terminal 3.5%EPS, terminal 4.0%EPS, terminal 4.5%EPS, terminal 5.0%
Q3 20275.62%$311.88$312.01$312.13$312.25$312.37
Q3 20285.33%$327.71$328.23$328.75$329.27$329.79
Q3 20295.03%$342.84$344.06$345.29$346.52$347.76
Q3 20304.74%$357.07$359.36$361.65$363.95$366.26
Q3 20314.44%$370.26$373.97$377.71$381.48$385.28
Q3 20324.15%$382.22$387.77$393.38$399.05$404.79
Q3 2033after the fade4.00%$393.69$401.34$409.11$417.01$425.03
Growth over the fade+29.3%+31.2%+33.1%+35.0%+37.0%
About this exhibit
Table: trailing reported EPS each year to the end of the path, grown along the two-stage model's fade from near-term growth to each terminal rate around the assumed one; model-conditional.
The earnings path the assumed rate implies over the fade (model-conditional)
Trailing reported EPS from Q3 2026, grown along the two-stage model's 6-year fade to terminal growth, at the assumed rate and ±0.5 and ±1.0.
Model-conditional, not a forecast. Growth falls in a straight line from the Valuation page's near-term 5.77% to terminal growth over 6 years, then runs at that rate; the bold column is the assumed rate. The earnings path: growth falls linearly from g_S to g_L over the 6-year fade, year t's rate is the average of that line over the year, g_S + (g_L − g_S)(t − ½) ÷ 6, and the year after grows at g_L. The growth above g_L summed over the fade is H(g_S − g_L), the identity's own term — a check. It starts from Shiller's trailing reported EPS ($295.52, Q3 2026); the Valuation identity prices trailing earnings, so the path is reported earnings, not FactSet's operating estimates.
Source: Robert J. Shiller; the Valuation page; Macro Review calculations · data to 16 Nov 2026cross.longrun-8

Appendix

How the statements are specified and estimated, how the fade reaches its terminal state, what the record scores, and every source209 checks, all pass
About this page

The S&P 500's own financial statements, read through the fund that holds it and carried forward: revenue from nominal GDP, the dollar and oil; a pretax margin that reverts toward its mean; taxes; dividends by Lintner's rule, buybacks and issuance to the share count; book value by clean surplus, so the return on equity is an output. Eight quarters explicit, then a fade to a long-run state the payout identity closes. Beside it, what the path assumes about the ten largest companies, how the model has done at each quarter since 2021 on the filings as first reported, and the other readings of the same earnings. Everything forward here is model-conditional: what the statements imply if the equations hold, not a forecast of what will be filed.

The statements, eight quarters explicit

Δln⁡St=βg Δln⁡NGDPt+β$ Δln⁡USDt+βo Δln⁡OILt+δ\Delta\ln S_t = \beta_g\,\Delta\ln \mathrm{NGDP}_t + \beta_{\$}\,\Delta\ln \mathrm{USD}_t + \beta_o\,\Delta\ln \mathrm{OIL}_t + \delta
ln⁡mt=μ+ϕ (ln⁡mt−1−μ)+γ Δln⁡St,NIt=St mt (1−τ) o\ln m_t = \mu + \phi\,(\ln m_{t-1} - \mu) + \gamma\,\Delta\ln S_t, \qquad NI_t = S_t\,m_t\,(1-\tau)\,o
Dt=Dt−1+c (d∗NIt−Dt−1),BBt=bt NIt,ISSt=ι4 MktCapt−1D_t = D_{t-1} + c\,(d^{*} NI_t - D_{t-1}), \quad BB_t = b_t\,NI_t, \quad ISS_t = \tfrac{\iota}{4}\,\mathrm{MktCap}_{t-1}
Bt=Bt−1+NIt−Dt−BBt+ISSt+Xt,Nt=Nt−1(1−BBt−ISStMktCapt−1+ν4)B_t = B_{t-1} + NI_t - D_t - BB_t + ISS_t + X_t, \quad N_t = N_{t-1}\Big(1 - \tfrac{BB_t - ISS_t}{\mathrm{MktCap}_{t-1}} + \tfrac{\nu}{4}\Big)
SymbolMeaningFromAssumed or estimated
SS, mm, NINIrevenue, the pretax margin, net income (per index unit at the start)the index basis–
DD, BBBB, ISSISS, XX, BB, NNdividends, buybacks, issuance, the clean-surplus residual, book value, the share countthe index basis–
βg\beta_grevenue's elasticity to nominal GDPNIPA Table 1.14, shrunk toward one–
cc, d∗d^{*}Lintner's adjustment speed and target payoutShiller's D and E since 1990–

Model-conditional, not a forecast. Every identity of Appendix A §A.3 holds on every row to 1e−9.

The fade and the terminal state

ROEyo=NIy/By−1→ROE∗,gS,y→gL,ln⁡my→μ\mathrm{ROE}^{o}_y = NI_y / B_{y-1} \to \mathrm{ROE}^{*}, \qquad g_{S,y} \to g_L, \qquad \ln m_y \to \mu
kT≡D+BB−ISS−XNI=1−gLROE∗,gps=(1+gL) NˉT/NˉT+1−1k_T \equiv \frac{D + BB - ISS - X}{NI} = 1 - \frac{g_L}{\mathrm{ROE}^{*}}, \qquad g_{ps} = (1+g_L)\,\bar N_T/\bar N_{T+1} - 1
SymbolMeaningFromAssumed or estimated
gLg_Llong-run nominal growththe SEP's longer-run mediansassumed
ROE∗\mathrm{ROE}^{*}long-run return on equity, opening bookthe index basis's mean (VP6)assumed

The terminal year's book grows at g_L and every flow after it too; the valuation forms read the path to year ten and this state after.

Exhibit 22 · Specification

How the numbers are made, step by step

  1. The index basis. Source: the members' filings, SPY's N-PORT holdings, the index's close. Transform: each quarter per index unit; the start quarter. Check: coverage at the start; the two buckets add to the index (4 checks, all passing).
  2. The parameters. Source: step 1, NIPA Tables 1.14 and 1.1.5, Shiller, H.10, EIA, the SEP. Transform: each parameter with its error, window and source. Check: β_g against statsmodels and its shrinkage; the margin filter reproduces its likelihood (5 checks, all passing).
  3. The path and the band. Source: steps 1–2 and the drivers. Transform: eight quarters, the fade, the terminal; the draws widened by the long record. Check: every identity on every row of the base and the alternatives; the quantiles from the draws (17 checks, all passing).
  4. The record. Source: steps 1–3 at each quarter since 2021. Transform: errors, benchmarks, PIT and coverage; the national accounts since 1970. Check: the summary recomputes from its rows (1 check, all passing).
  5. The references. Source: the GDP nowcast, the Payout, Earnings and Valuation pages. Transform: the nowcast, the share-count effect, g_L, the other legs. Check: each copy equals its source (no check guards this step).
  • Basis
    SPY's holdings, the index's own figures.

    Model-conditional: the paths on this page are what the index's own statements imply, carried forward by the equations below, if the equations are right — not a forecast of what will be filed. Every line is on SPY's basis: each member's per-share figure (its total over its diluted count for the quarter) times the shares the fund holds, over the value of those holdings, times the S&P 500's close — the index's own figure, through the fund (the author's decision of 12 Oct 2026 (D-2026-10-12-VP1)). The share count at the start quarter's end is one, so an aggregate is per index unit at the start.

  • Explicit
    Eight quarters of statements.

    Eight quarters from Q2 2026: revenue grows with nominal GDP (β_g 1.208), the broad dollar (β_$ 0.020) and WTI (β_o 0.059) plus a drift of 0.17% a quarter; the pretax margin's log gap to its mean (13.3%) shrinks by 16% a quarter and moves 0.60 per point of revenue growth; taxes take 19.3% and 98.9% of what is left reaches shareholders. Dividends adjust 2.6% of the way to 61% of earnings each quarter; buybacks start at 39.4% of earnings and fade over the ten years to the terminal split; issuance is 0.15% of market value a year; the share count falls by net buybacks over market value and rises by −0.18% a year of dilution; book value rolls forward by clean surplus with a residual of 0.30% of book a year. The market value holds the start's multiple (a constant price is the sensitivity).

  • Fade
    Years 3–10 to the long run.

    Years 3–10: revenue growth fades linearly from the explicit stage's exit rate (5.95% a year) to g_L = 4.0% (the SEP's longer-run real GDP growth plus PCE inflation); the margin completes its reversion to its mean by the year before the terminal year; return on equity on opening book fades linearly from year 3's 18.9% to ROE* = 18.8% (the index basis's mean since Q3 2020); the payout is what clean surplus leaves, and at the terminal year it is 1 − g_L ÷ ROE* = 78.7% net of issuance and the residual. As first specified, net income growth itself faded from the year-2 rate; with the explicit stage exiting while the margin still reverts, that carried a margin decline into years 3–10, so net income follows revenue and the margin here instead.

  • Estimation
    Each parameter's window and source.

    β_g on NIPA Table 1.14 (nonfinancial corporate gross value added on GDP, Q2 1947 – Q2 2026), estimated 1.256 (se 0.048), shrunk toward 1 with a prior sd of 0.10. δ, β_$ and β_o on four-quarter changes of the index basis's revenue (aggregate: per unit plus the measured share-count effect) with company totals before, stated. The margin's φ and γ by maximum likelihood with a measurement equation for the filings' one-quarter noise, Q1 2017 – Q2 2026, 4 pandemic quarters out; with the pandemic's quarters in, φ 0.676. Lintner's c and d* on Shiller's dividends and earnings since 1990, the index basis as the check. The error scales come from the long record (Table 1.14), never from in-sample residuals; the band adds the filings' noise and is widened at each horizon by the factor the margin equation needed on the national accounts since 1970 (1.23, 1.74, 2.26 at 1, 4 and 8 quarters).

  • Buckets
    The ten largest and the rest.

    The ten largest companies by the fund's value on the start quarter's report date — a fixed set over the horizon — and the rest each run the same equations with the index's dynamics (β, δ, φ, γ, c, ι, ν, θ, λ) and their own levels (μ, d*, x̄, τ, o, ROE*, the payout split) and starting values (the author's decision of 12 Oct 2026 (D-2026-10-12-VP16)). The published path is the single aggregate run; the buckets' sum beside it differs by the aggregation effect; the aggregate less the rest is what the index path assumes about the ten. No company has a driver of its own.

  • Record
    At each quarter since 2021.

    At every quarter from Q2 2021: the model started from the index basis as filed by then (each quarter as first reported) and run on nominal GDP, the dollar and oil as they came — conditional on the macro path, as published now; scored at 1, 4 and 8 quarters against the trailing earnings then filed, beside last year's growth carried forward and Shiller's trend (the log-linear trend of his E over the thirty years to the origin). The parameters are today's fit on the whole sample: the data are point in time, the parameters are not. Windows overlap beyond one quarter, so the independence test runs on the non-overlapping subsequence.

  • Switching
    When the base becomes the bridge's growth.

    The structural leg replaces the statistical one as the near-term growth in the Valuation page's two-stage bridge when its eight-quarter error is lower over the newest 12 origins of the paths as published, consecutive quarters (the author's decision of 12 Oct 2026 (D-2026-10-12-VP13)). Today: 0 published paths have their eight-quarter outcome; the rule is not met.

  • Legs
    The other readings beside the base.

    Three legs, each computed here: the structural (this page), the statistical (the Earnings page's growth process, on Shiller's basis) and the market-implied (the long-run growth today's price implies on the Valuation page). FactSet's bottom-up consensus (operating earnings) and the Livingston Survey's profits forecast (national accounts) are dated readings beside them, never inputs (the author's decision of 12 Oct 2026 (D-2026-10-12-VP3)); a reading past its age turns its own row into a dated notice and blocks nothing (the author's decision of 12 Oct 2026 (D-2026-10-12-VP12)).

Source: U.S. Securities and Exchange Commission (EDGAR: the members' filings; Form N-PORT); State Street Global Advisors; Yahoo Finance; BEA; Robert J. Shiller; Federal Reserve Board; EIA; Macro Review calculationscross.earnings_outlook-1

Before you read

Macro Review is a personal research site, still being built.

Nothing on it is investment advice, a recommendation or a solicitation.

Check every figure for yourself. Each names its publisher — confirm it there before you rely on it.

Figures can be revised, delayed or wrong.

Cones, fans, scenarios, fair values and “what followed” tables are conditional on the model each page names: they describe, they do not predict.

No warranty is given; use is at your own risk.