The ‘haunted’ housing market’s real problem may be too few people, not too few homes, says J.P. Morgan’s David Kelly

J.P. Morgan Asset Management strategist David Kelly expects U.S. population growth of about 800,000 people annually over the next few years, implying demographic demand for roughly 320,000 new homes per year, versus about 1 million annually in the first two decades of this century. Rental vacancies rose to 7.6% in Q3, buyer traffic fell to 18 in October, and Kelly forecasts housing starts of 1.32 million this year and 1.24 million in 2027. He cites 7.32% 30-year mortgage rates, tighter immigration, tariffs and competition from data-center construction as headwinds; data-center spending rose 73% year over year through August to an $85 billion annual rate.

Analysis

The market may be conflating a long-run affordability/availability problem with near-term demand for new construction. If household formation slows while recent completions work through the system, builders face weaker absorption and more discounting; lower starts then become a lagged support for supply and pricing, not an immediate demand catalyst. This favors a selective rather than blanket housing short: large builders may gain share from smaller or less-capitalized peers, but share gains will not protect margins if incentives and labor/material costs keep rising.

The AI buildout is a second-order negative for residential construction: it competes for trades and materials, potentially preventing lower starts from translating quickly into lower construction costs. Data-center developers and contractors could also see cost pressure, so the labor squeeze is not an uncomplicated AI-beneficiary signal. Rental owners face near-term leasing risk where vacancy is rising; declining starts could help later, but geography and unit type matter.

Near term, housing-sensitive equities remain exposed to traffic, cancellations, incentives, and mortgage-rate volatility. Over 1–3 months, watch starts, builder confidence, vacancy and wage/inflation data; higher wage growth or inflation expectations could keep rates restrictive. Over 6–18 months, the key uncertainty is whether weaker household growth persists or affordability-driven household formation rebounds. The contrarian risk to a bearish view: vacancy measures and aggregate household projections can miss local shortages, and constrained affordability may represent deferred—not extinguished—demand. A sharp rate decline could unlock that demand quickly.

Key Decisions for Investors

• Tactical bias: underweight a broad homebuilder basket versus the broad equity market rather than initiate an aggressive outright short. Entry is more attractive after any rally unsupported by improving buyer traffic or lower incentives; keep risk limited because rate relief can trigger a fast rebound.
• Prefer larger, better-capitalized builders over smaller builders in any housing exposure, but verify relative valuation, cancellation trends, and incentive intensity before expressing the pair. The thesis is share consolidation, not immunity from sector-wide margin pressure.
• Avoid treating rising rental vacancy as a clean short across all residential property. Watch local rent growth, concessions, and new supply by market; consider reducing exposure only where leasing metrics deteriorate alongside elevated deliveries.
• Falsify the cautious view if builder traffic and starts recover for multiple readings, incentives retreat, and rental vacancy turns down. Escalate the bearish view if starts weaken while price cuts, cancellations, or construction-wage pressure accelerate; reassess rates if wage growth or inflation expectations move materially higher.