Durable Goods Orders vs S&P 500

The Census Bureau publishes two durable goods releases each month: the advance release (DGORDER on FRED) typically four weeks after the reference month, and the final release roughly one week later as part of the broader M3 manufacturers' shipments, inventories, and orders report. The series covers approximately 92 industries grouped into seven categories, with transportation equipment representing the most volatile component because Boeing widebody orders and military aircraft contracts arrive in lumps of $5 to $40 billion in single months. The Census Bureau's seasonal adjustment treats transportation as a separate category specifically because of this lumpiness, and most professional readers ignore the headline DGORDER print in favor of the ex-transportation series. The advance release covers approximately 65 percent of the final M3 universe, with the remainder added in the subsequent full release.

The relevant policy desk is the Federal Reserve's Industrial Production team within the Board's Division of Research and Statistics, which uses M3 nondefense capital goods ex-aircraft (the 'core capex' series, FRED:NEWORDER) as the primary input to its quarterly equipment investment forecast. The pair captures the difference between equity-market price action on the S&P 500 and the order-flow data that ultimately drives the equipment-investment line in nominal GDP. When SPY rallies without core capex support, the rally is being carried by multiple expansion or services-sector earnings rather than by the manufacturing-capex cycle that the durable goods report tracks. The Atlanta Fed GDPNow model gives durable goods a roughly 12 percent weight in the equipment-investment subcomponent of its quarterly nowcast, which is why a single durable goods print can move the GDPNow estimate by 10 to 30 basis points.

Boeing's order book alone has produced four single-month durable goods swings of more than 20 percent since 2010: the August 2014 +22.6 percent print (Boeing booked 324 aircraft at the Farnborough air show), the July 2014 -18.4 percent print (post-Farnborough reversion), the April 2017 -2.1 percent print (Boeing 737 MAX disclosure), and the December 2024 +9.2 percent print (year-end catch-up bookings after the IAM strike resolved). Each of these moves was the headline number, and each of them was almost entirely Boeing-specific. The ex-transportation series tracked through all four episodes within a +/- 0.5 percent band, and the ex-defense series within a +/- 0.7 percent band. The implication is that the headline DGORDER print in any given month reveals more about Boeing's order book than about the broader US capex cycle.

The core capex series, nondefense capital goods orders ex-aircraft, is the cleaner proxy for business investment intent. The series typically runs at $73 to $76 billion per month in 2026 dollars, has a three-month moving-average correlation with quarterly equipment investment in real GDP of +0.71 since 1992, and leads the equipment investment line by approximately one to two quarters. The cleanest near-term signal in the March 2026 release was not the headline +0.8 percent but the +0.4 percent core capex print, which marked the third consecutive month above $74 billion, the longest such streak since the AI-capex acceleration began in mid-2023. Defense capital goods orders, the third leg, run at a much lower base rate ($14-17 billion per month) but produce occasional 30 percent monthly swings on lumpy contract awards that should always be filtered before drawing capex-cycle conclusions.

The cleanest stress test on the durable orders versus SPY relationship is the 2008-2009 collapse. Headline DGORDER fell from $232.5 billion in July 2008 to $158.2 billion in March 2009, a peak-to-trough decline of -32.0 percent. Ex-transportation orders fell from $159.5 billion to $115.4 billion over the same window, a -27.7 percent decline. Core capex orders fell from $74.1 billion to $51.8 billion, a -30.1 percent decline. SPY fell from $146.40 on October 9, 2007 to $67.10 on March 9, 2009, a -54.2 percent decline. The percentage decline in core capex was almost exactly half the percentage decline in SPY, an unusually clean ratio that has held in subsequent recessions.

The calibration point is the lead-lag: the SPY peak preceded the durable orders peak by nine months, but the SPY trough preceded the durable orders trough by zero days. SPY bottomed on March 9, 2009; the M3 release for March 2009 was published on April 24, 2009, marking the same month as the equity low. The asymmetry is structural: equity markets price discount-rate inflections within hours, while order books require firms to actually place new contracts, which lags the trough by exactly the time it takes to move from board-level approval to executed PO. The 2020 episode replayed the same pattern in compressed form: SPY trough on March 23, 2020, and durable goods trough in April 2020, separated by approximately five weeks. The 2001 dot-com episode took longer to resolve, with SPY bottoming on October 9, 2002 and core capex bottoming in February 2003, a four-month gap that reflected the slower industrial recovery from the equipment overhang of the late-1990s capex boom.

The Convex Net Liquidity Impulse (CNLI) treats core capex orders as one of the five real-economy confirmation channels for the liquidity-to-output transmission. The current setup, with core capex above $74 billion for three consecutive months and SPY near record highs, registers as a confirming rather than diverging signal: the equity rally is being supported by underlying capex acceleration rather than running ahead of it. The CNLI overlay flags this as a 'liquidity-and-throughput aligned' regime, the same configuration that prevailed from May 2017 through January 2018 and that supported the late-2017 SPY rally rather than warning against it. The June 2003 to October 2007 stretch produced the longest sustained version of the same alignment in the modern record.

The disagreement watch is on the headline number rather than the core. The headline DGORDER series printed three consecutive monthly declines through February 2026 before the +0.8 percent March rebound, which produced a brief but visible CNLI flag in February. The flag cleared with the March print because the underlying core series never confirmed the headline decline; the three-month decline was almost entirely Boeing 737 MAX delivery disruptions and a year-end defense-spending pullback. Disciplined CNLI reading separated those signals in real time and avoided the false negative that headline-only readers would have flagged. The November 2018 to January 2019 episode is the cleanest historical case where headline durable goods declined for three consecutive months while core capex held above its 12-month moving average; the divergence was correctly read as a transportation-and-defense lumpiness story rather than as a cycle peak.

Three episodes since 2000 illustrate the pair leading the broader cycle correctly. First, the August 2000 peak in core capex at $66.2 billion preceded the March 2001 NBER recession start by seven months, while SPY peaked one month later in September 2000 at $151. The pair flagged the cycle peak before either series alone provided a clean signal. Second, the May 2008 peak in core capex at $74.1 billion preceded the December 2007 NBER recession start by negative five months (NBER backdated the recession), but the May 2008 SPY level of $141 represented a 7 percent decline from the October 2007 peak, providing the cleanest cross-confirmation of cycle weakness. Third, the August 2019 peak in core capex at $69.2 billion was followed by a six-month plateau through January 2020, which corresponded to SPY's grind-higher rally before the COVID shock; the pair did not warn of COVID, but it also did not falsely flag a cycle peak that was not yet present.

The one episode where the pair failed was the 2015-2016 industrial recession. Core capex declined from $73.8 billion in November 2014 to $63.6 billion in April 2016, a -13.8 percent decline that did not produce an NBER recession because the weakness was concentrated in oil-and-gas and dollar-strength-exposed manufacturing while the broader services economy continued expanding. SPY drew down only -14.0 percent through February 2016 and recovered fully by July. The lesson is that core capex weakness without services confirmation is insufficient to call a broader cycle turn, which is why the pair is best read alongside ISM Services and BLS payrolls rather than alone.

The April 2026 advance M3 release is scheduled for May 27, 2026, and the May release for June 25, 2026. Three specific items in those releases will determine whether the current pair configuration tightens or diverges. First, the Boeing component: 737 MAX deliveries are running approximately 28 per month against a target of 38, and any acceleration above 35 in a single month would produce a positive headline distortion that the disciplined reader should ignore. Second, the AI-capex line: computers and electronic products orders rose +3.7 percent ($1.0 billion) in March 2026, the strongest single-month move since November 2023, and continuation of that trend would confirm the AI capex story that has driven both NVDA and core capex throughout 2024-2026. Third, the defense component: the FY2026 defense appropriation passed in March includes a $42 billion increase in procurement spending that will hit the M3 release on a four-to-six month lag.

The pair-watch threshold is core capex breaking below $73 billion for two consecutive months, which historically (2001, 2008, 2015, 2020) has marked the beginning of an industrial slowdown that takes nine to fifteen months to clear. Above $75 billion for two consecutive months would confirm continued capex acceleration. The current $74 billion run rate sits in the middle of that range, which is why the pair currently provides regime confirmation rather than a directional signal. The May 27 advance release will be the cleanest near-term observation point given the AI-capex momentum visible in the March print.