PulteGroup Q3 2026: $2.53 EPS Falls 14.5% as Closings Drop 9% and Margins Compress

PulteGroup (NYSE: PHM), the Atlanta homebuilder, reported third-quarter 2026 net income of $475.7 million, down 18.8% year over year, with diluted earnings per share of $2.53, down 14.5%. Home sale revenues fell 9.1% to $3,859.0 million on 9% fewer closings (6,876 homes at a $561,000 average price, down 1%), and home sale gross margin compressed to 24.8% from 26.2%. The culprit stack is familiar — rising mortgage rates, stretched affordability, sidelined buyers — but the quarter's real tension sits one line lower: net new orders fell only 6% while backlog grew 4% to 10,298 homes. Demand is delayed more than destroyed.

The margin line deserves a closer read: the 24.8% includes $26 million of land impairments (~70 basis points), so the clean margin is closer to 25.5% — still down 0.7 points, with SG&A deleveraging to 10.3% of home sale revenues from 9.4% doing the rest of the damage. When closings fall faster than overhead, every basis point of operating leverage runs backward.

The Headline Numbers​

Nine-month EPS of $6.79 is down 20.6% — a steeper decline than the quarter, which says Q3 was actually the less bad part of the year. Financial Services remains the stabilizer: pretax income flat at $46 million with the mortgage capture rate up to 85%, as the captive lender keeps financing the homes the market won't. Land sale revenue collapsed 85% to $7.8 million, but at 35% margins on scraps, it never mattered.

Operations: Selling Slower, Building Anyway​

The quarter's operating paradox: orders down 6%, community count up 7% to 1,077, backlog up 4%. PulteGroup opened communities into a soft market and banked the demand rather than chasing price — backlog value up 3% to $6.4 billion even as the average order price softened. That is the "balance the price and pace" discipline CEO Ryan Marshall describes, and it shows up on the balance sheet as inventory, not concessions: house and land inventory rose 9.0% year-to-date to $14,090.5 million while cash fell $600.7 million.

Customer deposits jumped 33.0% to $515.9 million — the cash signature of that growing backlog — while the average community count growth (+7%) against order declines (−6%) implies absorption per community fell roughly 12%. More stores, fewer sales each: the classic late-cycle homebuilder footprint, funded by a fortress balance sheet rather than leverage.

The Balance Sheet: $1.4B of Cash and 12% Debt-to-Capital​

Cash fell $600.7 million in nine months while inventory absorbed $1,165.1 million — the working-capital cycle of a builder that keeps starting homes into soft demand. Notes payable rose $177.8 million even as the debt-to-capital ratio holds at just 12%, and Financial Services debt fell $149.1 million as mortgage originations slowed. Equity still grew 0.8% despite $1.0 billion of nine-month buybacks, because $1,294.7 million of earnings outran the return.

The buyback cadence is the quiet bull case: 2.8 million shares at $124.70 in Q3, 8.4 million shares (4% of the float) at $123.51 over nine months. With the stock pricing distress and the balance sheet pricing strength, retiring 4% a year at ~11× depressed earnings is the capital allocation doing its job.

The Multi-Year Arc: Peak $14.69, Then the Rate Cycle​

Five years of ledger, one cycle: the post-pandemic boom peaked in FY2024 at $14.69 per share, then rates did their work — FY2025 net income fell 28.1% on a 3.5% revenue decline, as gross margins compressed and SG&A deleveraged. The share count tells the other half: 249.3 million shares in FY2021 down to 192.7 million in FY2025, a 23% shrinkage that turned a revenue decline into merely an earnings decline. Without buybacks, FY2025 EPS would print near $8.50 instead of $11.12.

Goodwill took a $28.6 million impairment in FY2025 (down to $40.4 million) and other income swung to a ($91.5) million charge — the clean-up items of a management team admitting the cycle turned. Q3 2026's $26 million of land impairments continue that honesty.

The One Big Question: Is Backlog Demand or Denial?​

Everything constructive in this quarter — backlog +4%, deposits +33%, community count +7% — assumes today's orders become tomorrow's closings at today's prices. The bull case: a 10,298-home backlog at $6.4 billion is ~1.5 quarters of closings already sold, mortgage rates eventually relent, and PulteGroup's 60%+ move-up and active-adult mix (buyers with equity, less rate-sensitive) converts better than starter-home peers. The impairments stay small because land was bought right.

The bear case: backlog grew because closings fell faster than orders — a 6,876-closing quarter against 6,208 orders means the company should have drawn backlog down, yet it grew, which implies cycle times stretching or buyers delaying. If rates stay high, today's $6.4 billion backlog becomes tomorrow's cancellations and incentives, and the $14.1 billion of inventory becomes the market's problem. The $26 million of impairments is either the beginning of honesty or the end of it — watch this line next quarter.

Tracking an $18.5 Billion Homebuilder Balance Sheet in Plain Text​

A homebuilder's books are dirt and lumber capitalized: $14,090.5 million of house and land inventory, $433.9 million of mortgages awaiting sale, $747.2 million owed to trades — and $13,096.0 million of equity, every dollar reconciling to $18,484.9 million of assets across 14 accounts. This is how we model every company. The Q3 income statement, exactly as filed (income negative, expenses positive, net income absorbed by Equity:Adjustments):

; Check: −3,973.846 + 2,906.902 + 61.468 + 397.198 + 2.772 − 19.700 + 149.479 + 475.727 = 0 ✓
2026-09-30 * "PulteGroup, Inc." "FY2026Q3 Income Statement"
  Income:Revenue                               -3973.846 MUSD  ; total revenues: homebuilding 3,866.820 + Financial Services 107.026
  Expenses:CostOfRevenue                        2906.902 MUSD  ; homebuilding cost of revenues — the filing's own line
  Expenses:OtherOperatingExpenses                 61.468 MUSD  ; Financial Services expenses — the filing's own line
  Expenses:SellingGeneralAdministrative          397.198 MUSD  ; selling, general, and administrative expenses — the filing's own line
  Income:OtherNet                                  2.772 MUSD  ; equity loss from unconsolidated entities, net — a charge
  Income:OtherNet                                -19.700 MUSD  ; other income, net — a credit
  Expenses:IncomeTax                             149.479 MUSD  ; income tax expense
  Equity:Adjustments                             475.727 MUSD  ; net income (surplus set by balance assertion; no noncontrolling interest)

The Verdict: Bull vs. Bear​

Bull Case

• Backlog +4% to 10,298 homes ($6.4B) with deposits +33% — forward revenue already sold.
• 60%+ move-up/active-adult mix converts through rate pressure better than starter peers.
• 12% debt-to-capital with $1.4B cash — the cycle can't force a mistake.
• 8.4M shares (4%) retired at $123.51 — buybacks compounding through the trough.
• 24.8% gross margin still best-in-class; impairments honest and small ($26M).
• FS pretax flat at $46M with 85% capture — the captive lender stabilizes closings.

Bear Case

• Closings −9% with absorption per community down ~12% — demand deteriorating, not pausing.
• Gross margin −1.4 pts with SG&A deleveraging to 10.3% — operating leverage in reverse.
• Backlog growing while closings fall smells of stretching cycle times, not strength.
• Inventory +9% YTD to $14.1B while cash −30% — working capital eating the fortress.
• 9M EPS −20.6% and accelerating impairments — the downcycle may have further to run.
• Land impairments ($26M) could be the first of many if prices roll over.