Should you buy selloff in defense stocks? Bernstein answers

Bernstein recommends waiting until after the November 3 midterm elections to buy U.S. defense stocks, which have moved from a 15% premium to the S&P 500 in February to a 12% discount amid a stalled federal budget. A continuing resolution through December 11 could hold funding at 2026 levels and pressure growth areas including tactical missiles and interceptors. Despite near-term caution, Bernstein expects the 2027 defense investment budget to increase by more than 10%, viewing the election as a potential clearing event for the sector.

Analysis

The defense selloff is primarily a duration-of-appropriations problem rather than a deterioration in end-demand: primes cannot convert elevated requirements into revenue, margin, or backlog de-risking while new-start authority and production-rate increases are constrained. The largest near-term exposure sits with missile, munitions, and space-defense programs, making RTX, LMT, NOC, LHX and GD more sensitive than diversified services names; a flat-funded environment also favors sustainment-heavy contractors such as LDOS and BAH relative to platform manufacturers. The second-order risk is that subscale solid-rocket-motor, electronics, and machining suppliers absorb the working-capital burden before primes do, potentially creating supplier distress and later delivery bottlenecks.

Over the next 1-3 months, the relevant catalyst is not geopolitical escalation but a credible appropriations path and evidence that production funding can be obligated rather than merely authorized. A post-election entry could be attractive if valuations remain depressed, but the proposed spending-growth outcome is still an analyst forecast rather than a funded program; a prolonged stopgap would push revenue recognition and operating leverage into later fiscal periods, warranting lower near-term EPS estimates. For 6-18 months, depleted inventories and replenishment requirements support a constructive setup, but only if contract awards, missile lot sizes, and cash-conversion guidance begin to validate it.

Consensus may be over-penalizing the group as a single macro basket. The better expression is to own contractors with recurring sustainment, classified, and international demand while avoiding names whose valuation requires immediate U.S. procurement acceleration; this reduces exposure to legislative timing without abandoning the structural replenishment thesis. AB has no direct operating sensitivity to defense procurement, so the item is not a fundamental catalyst for the asset manager.

Key Decisions for Investors

• No immediate broad defense-beta purchase before the election/appropriations clearing event; maintain a watchlist rather than chase a headline-driven bounce in ITA or XAR. Reassess within 5 trading days of a funding resolution, with contract-award visibility as the required confirmation.
• Prepare a 6-12 month pair trade: long LDOS or BAH / short ITA in equal beta-adjusted dollars. Sustainment and services revenue should be relatively resilient under constrained new obligations; exit if a full-year appropriations bill materially accelerates platform and missile procurement, which would favor the broader prime basket.
• For post-election deployment, favor a staged long in RTX and LHX over LMT/NOC: initiate one-third only after appropriations language preserves missile, air-defense, and electronics production growth; add on confirmed lot-size awards or raised free-cash-flow guidance. Thesis is falsified by a continuing-resolution extension that delays production increases beyond the next fiscal-year planning cycle.
• Monitor suppliers and program disclosures rather than conflict headlines: delayed awards, inventory build, or unfavorable cash-conversion commentary at RTX, LMT, NOC, and GD would signal that legislative friction is reaching the supply chain and should delay exposure.
• Do not position in AB on this development; any trade would require separate evidence of defense-sector AUM flows, performance-fee exposure, or net-new mandates.