Will the US economic slowdown worsen into a recession?

Introduction

The US real GDP growth rate in the fourth quarter of 2025 was only 1.4%, a sharp decline from 4.4% in the third quarter. This figure was far below market expectations and triggered widespread concerns about an economic slowdown. According to the preliminary estimate released by the US Bureau of Economic Analysis (BEA) in February 2026, the full-year GDP growth rate is 2.2%, a slowdown from 2.8% in 2024. This decline reflects the combined effects of multiple factors, including continued downward revisions to employment data, reduced spending due to the government shutdown, slower immigration inflows, and the potential peak in artificial intelligence (AI) capital spending. The Atlanta Fed's GDPNow model shows that the growth rate in the first quarter of 2026 is expected to be 3.1%, but this rebound may be driven by a short-term bounce rather than structural improvement.

At the same time, employment data reveals deeper structural problems.

In its January 2026 baseline revision, the Bureau of Labor Statistics (BLS) lowered its 2025 nonfarm payroll growth estimate from 584,000 to 181,000, a net decrease of 403,000. This revision, based on the Quarterly Employment and Wage Census (QCEW) data, shows that the deviation between the estimated and actual nonfarm payrolls (NFP) has reached historical extremes, with a deviation of four standard deviations in 2024 and eight standard deviations in 2025. This systematic error casts serious doubt on the reliability of NFP data and suggests that the middle class has been in a state of implicit recession for two years. This article analyzes the reasons for the slowdown in US economic growth, potential risks, and the possibility of a recession in 2026, based on the latest economic data and market dynamics. It focuses on the impact of distorted employment data, peak AI investment, pressures in the housing and private credit markets, bond market signals, and the Federal Reserve's policy shift. By integrating the Conference Board Leading Economic Index (LEI), Consumer Confidence Index, and other key indicators, this comprehensive assessment aims to provide a reference for investors and policymakers. Employment Data Distortion and Fading Economic Support Factors Employment data has always been a core indicator for assessing economic health, but the reliability of NFP estimates has declined significantly in recent years. The NFP is based on a monthly survey of approximately 600,000 businesses, supplemented by a birth-death model adjustment, but the QCEW, as "real" data based on actual reports from 16 million businesses, often shows significant bias. The 2025 revision shows only 181,000 new jobs for the year, far below the initial report, a cumulative downward revision of 898,000. This bias not only reflects statistical methodological problems but may also involve bureaucratic inefficiency, fraud, or data manipulation for national security reasons. This distortion directly impacts the Federal Reserve's decisions. NFP data, once considered evidence of a strong economy, has been revised to show that job growth for 2023-2025 was overestimated by 1-2 million. Atlanta Fed President Bostic acknowledged in a February 2026 speech that the revised data "adjusted the 2025 employment picture," suggesting the labor market is not as strong as it appears. In January 2026, nonfarm payrolls increased by 130,000 and the unemployment rate fell to 4.3%, but initial jobless claims fell to 220,000, indicating a still tight labor market accompanied by slowing wage growth. Another key factor in the slowdown in economic growth is the fading of supporting forces. From 2023-2025, illegal immigration inflows supported about half of economic growth, sustaining expansion by increasing labor supply and consumer demand. However, stricter border controls in early 2025 and the exposure of immigration payment and fraud issues caused this "almost-in-the-room" factor to subside. While the government shutdown in 2025 was brief, it dragged down fourth-quarter GDP growth by 0.9 percentage points, highlighting the impact of fiscal uncertainty on the economy. Furthermore, AI capital expenditure was a significant driver of GDP growth in 2025, contributing about half of the increase. However, data from February 2026 suggests that this expenditure may have peaked. The widening spread in credit default swaps (CDS) between Oracle and CoreWeave reflects market skepticism about the sustainability of AI infrastructure investment. AI capital expenditure is projected to reach $650-700 billion in 2026, primarily driven by giants such as Microsoft, Amazon, Alphabet, and Meta, but this scale could lead to cash flow pressure. Evercore ISI analysts warned that increased capital expenditure could cause some companies to turn negative cash flow, triggering a "red flag" signal. Market Reaction and Sector Rotation Despite weak economic data, the stock market has remained relatively stable. The S&P 500 (SPX) has been consolidating since October 2025, with current prices comparable to then and only 2% away from its all-time high. This apparent indifference reflects the market's ignoring of recession signals, but internal cracks are emerging. AI-related stocks account for 35% of the SPX's market capitalization, but valuations have reached dot-com era levels, suggesting a potential zero return over ten years, including dividends. Sector rotation is shifting from technology to cyclical stocks, a typical sign of the end of a cycle, but is expected to be short-lived. The bond market is more sensitive to growth expectations. In February 2026, the 10-year Treasury yield remained stable at 4.05%, down 0.17 percentage points from the previous month, indicating a decline in inflation and growth expectations. The 30-year yield fell to 4.70%, and the 2-year yield to 3.44%. This steepening yield curve suggests that the market expects a Federal Reserve rate cut, but not an immediate one. The FOMC's LEI fell 0.2% in December 2025, marking its fifth consecutive month of decline, signaling economic weakness at the beginning of 2026. The Consumer Confidence Index rose to 91.2, but the short-term expectations sub-index was 72, remaining below the 80 threshold for 13 consecutive months, suggesting a risk of recession. The housing market accounts for 20% of the US economy and is currently showing significant signs of weakness. Rents plummeted in 2025, with the national median rent falling to $1,693, a 1% year-over-year decrease. Apartment List data shows the median rent is $1,367, down 1.1% year-over-year. Lower rents make renting more affordable than buying, and historical experience suggests this will lead to lower home prices. Zillow economists predict a 1.2% increase in home prices in 2026, Realtor.com projects 2.2%, but J.P. Morgan Global Research believes national home prices will remain flat at 0%. Residential investment is expected to decline quarter by quarter in 2025, falling 1.5% in the fourth quarter. The private credit market is facing increased risks. Over the past two years, commercial bank lending growth has primarily flowed into private credit and private equity, rather than the productive sector. This opaque, illiquid market is facing liquidity problems. The Blue Owl fund's closure of redemption channels indicates a liquidity crunch. The credit cycle has entered a Ponzi scheme phase, with loans based on asset price increases rather than cash flow serving debt. Default rates are expected to rise in 2026, with leveraged borrowers' refinancing needs peaking in 2028, relying on the private credit and syndicated loan markets. This phase is similar to the dot-com bubble, with junk bonds flowing to unrevenue-generating telecom companies. Currently, AI companies' infrastructure investments rely on equity and private credit, but revenue is insufficient to servic the debt. Market doubts about sustainability have led to widening CDS spreads. Moody's 2026 Private Credit Outlook points out that increased interconnectivity between funds and traditional financial institutions could amplify contagion risks. The expanding role of retail investors may increase volatility. AI Bubble Risks and Historical Comparisons The AI ​​investment boom is similar to the dot-com bubble. AI capital expenditure will support economic growth in 2025, but spending will peak in 2026, with the second derivative turning negative. Valuations are overvalued, similar to the railroad era or the dot-com bubble: initial capital evaporates, but infrastructure ultimately benefits. AI is a real technology, but overhyped and has not yet generated significant productivity gains. Corporate experiments have yielded no returns, and bottlenecks are hindering white-collar employment. Unlike dot-coms, this round of financing is from private credit rather than junk bonds, and opaque transactions increase risk. The future availability of infrastructure (such as Nvidia chips) is uncertain. Cisco has sold routers to companies with no revenue, similar to Nvidia's current ecosystem. Nvidia is expected to miss quarterly expectations, triggering a correction. Investors should increase their cash holdings and shift to bonds, especially 20-30 year Treasury bonds. BlackRock underweighted US duration assets at the beginning of the year, which is expected to be a misjudgment. The best trade in 2026 may be long-term Treasury bonds, benefiting from declining inflation and growth expectations. The Federal Reserve will follow the downward trend in interest rates, with current policy tightening at 1.5-2%.

Federal Reserve Policy Shift and Data Reliability

In May 2026, Kevin Warsh will succeed as Chairman of the Federal Reserve. He is considered a hawk, criticizing quantitative easing (QE) for expanding the balance sheet. Warsh advocates reducing the Fed's footprint in the economy and may coordinate with Treasury Secretary Scott Bessent to gradually reduce holdings of mortgage-backed securities (MBS) and shift towards short-term Treasury bonds. Policy is expected to be more flexible in 2026, with less forward guidance and an emphasis on data reliance.

Warsh may shift to private data such as ADP to avoid NFP bias. The market reacted with shock to the nomination, but PolyMarket shows he has consistently been a leading candidate. Policy focus: balancing inflation control and growth; the AI ​​productivity narrative influencing decision-making.

2026 Recession Probability and Key Signals

The NBER typically declares recessions after the fact, only confirming them after the 2007 stock market peak. Current indicators are mixed: LEI is weak, consumer confidence is below the threshold, but GDPNow Q1 is 3.1%, and the unemployment rate is 4.3%. The Richmond Fed SOS indicator shows that the average increase in the insured unemployment rate of 0.2 percentage points has not been triggered.

The probability of a recession is approximately 66%. If it occurs, the stock market will fall by 30-50%, and bonds will rise by 30%. The government may not acknowledge it in time, but portfolios will not be affected by the label. Key monitoring: 10-year, 30-year, and 3-month Treasury yields; credit market liquidity; AI CDS spreads.

Conclusion The US economic slowdown stems from data distortion, fading supporting factors, and asset bubble risks. In 2026, the risk of recession increases, but a shift in the Federal Reserve and fiscal stimulus can cushion the blow. Investors should pay attention to bond market signals and diversify their risk. Economic cycles are inevitable, but adequate preparation can capture opportunities. Future research focus: Emerging opportunities amidst the AI ​​wreckage.