S&P Global Ratings Economics' recession model implies the probability of a recession starting within the next 12 months is 30%-35% (see chart 1). Our model is based on key forward-looking financial market indicators and the Organization for Economic Co-operation and Development's leading economic index through August. While the probability of a recession has eased since spring, it remains elevated.

Chart 1

The Yield Curve Throws A Red Flag

The primary factor in the evolution of recession probability in our model is the yield curve. The spread between yields of 10-year and three-month Treasuries has been inverted since October of last year in this expansion (see chart 2). In May, the long end started to rise, thus reducing some of the inversion's depth. Since mid-May, 75% of the 80-basis-point rise has stemmed from a rise in real interest rates, and the other 25% owes to a boost in inflation expectations.

An inverted yield curve (in the U.S., though not necessarily other advanced countries) has predicted seven of the past seven recessions, with no false positives (unlike the stock market, which is well known for its false positives). A few times when the yield spread came close to inverting (for example, in 1995 and 1998), the Federal Reserve responded by preemptively cutting rates, which in hindsight appears to have helped extend the expansion. Historically, the time between the first month of an inversion and the start of a recession has ranged from six to 18 months.

Chart 2

The explanations for such tight correlation vary from financial market participants' expectations for central bank policy (linked to expectations of business cycle outcomes) to market attitudes toward various risks (also influenced by economic outcomes). Still, at its core, the current yield curve inversion primarily reflects that investors expect the Fed will eventually have to cut interest rates, which in turn may i) reflect the view that the Fed will go too far and push the economy into recession, or ii) simply reflect that the Fed has pushed the policy rate above its expected long-run average and that its success in bringing down inflation will allow it to later reverse course. We think the latter may be the case.

Besides, it may also be the case that yield curves are now more susceptible to inversions, given long-term natural interest rates have trended down in the past 40 years to under 1%. Over the years, as the Fed has become more transparent and large-scale asset purchases (quantitative easing) have become part of the norm, term premium on 10-year bond yields has become structurally low, which has also made yield spreads susceptible to inversions.

All told, while our econometric model suggests a 31.5% risk of a recession, down from spring, we remain wary considering it was flashing near 50-50 odds in May. Our dashboard of leading indicators is not quite indicating that the coast is clear, and coincident indicators are either in a late cycle or approaching one. Although our baseline economic forecast avoids recession in the next 12 months, the balance of risks is clearly tilted to the downside.

Mixed Signals On Momentum

Data for the second and third quarters so far demonstrates economic resilience, with growth surprising to the upside. The running estimate of real GDP growth for the third quarter is near 4%, an acceleration from the first half's 2.1% average.

Still, our dashboard of 10 principal forward-looking indicators gives mixed signals on near-term economic growth prospects (see below). As of August, four of the indicators were flashing negative signals on growth momentum, while the other six were neutral or positive. August data on the Freight Transportation Services Index is pending as of Sept. 20 {see the appendix for our rules on signal transition).

Elements reflecting expectations were split. The inverted term spread and equity market flashed negative in August as the S&P 500 index gyrated down from the high of 4589 on Aug. 1. On the other hand, the other leading indicator from the capital markets, the credit spread, turned positive last month, with both investment-grade and speculative-grade corporate credit spreads tightening further during the summer (see chart 3).

Chart 3

Consumer sentiment

The Consumer Sentiment Index from the University of Michigan rose again in August before inching down in the first half of September (see chart 4). The three-month moving average of this index also moved higher.

Chart 4

That said, we also look at the Conference Board's Consumer Confidence Index. The Conference Board's expectations index--based on consumers' short-term outlook for income, business, and labor market conditions--in August was barely above 80, the level that historically signals a recession within the next year.

Gas prices have also risen again since August, and that's surely going to erode some optimism. Taking these factors together, we think consumer are likely to become more careful with their wallets.

Job market

All the while, layoffs--as indicated by initial jobless claims--remain near historical lows. That said, hiring of temporary help services (a lesser-known leading indicator) has started to decline, which indicates that the pace of payroll job gains is likely to continue softening in the coming months (see charts 5 and 6).

Chart 5

Chart 6

Manufacturing

The Institute of Supply Management Manufacturing New Orders Index stayed below 50 in August, reflecting continued headwinds to growth from the manufacturing sector (see chart 7). Sentiment in the manufacturing sector turned sour 12 months ago and has yet to reverse. That said, a New Orders Index below 52.5 over time is generally consistent with a decrease in the Census Bureau's series on manufacturing orders.

On the other hand, Manufacturers New Orders: Nondefense Excluding Aircrafts has been moving up year to date (see chart 8). The Freight Transportation Services Index (data is pending for August), covering all domestic transport of commercial freight, turned neutral (from negative) in July.

Chart 7

Chart 8

Housing

Meanwhile, the housing sector appears to have found a floor. Single-family starts and, more important, permits (which offer a way to track future construction) have bounced up from January lows and revealed a grinding-up pattern in the past few months (see chart 9). Nonetheless, high interest rates will limit the positive impulse to growth in the housing market.

Chart 9

Bank lending and financial conditions

Banks are continuing to tighten lending standards across the board (see chart 10). Loan officers believe high inflation and interest rates are working to lower borrowers' debt servicing capacity, increasing their exposure to risk. Demand has continued to weaken across loan categories.

For the second half of 2023, banks have reported expecting to further tighten standards on all loan categories. They most frequently cited a less favorable or more uncertain economic outlook, as well as expected deterioration in collateral values and the credit quality of loans, as reasons for expecting to tighten lending standards.

Meanwhile, the Chicago Fed's National Financial Conditions Index (NFCI), which is a composite measure used to sense stress in the financial system, remains a tad below zero (even as the leverage subcomponent is elevated), which we consider neutral for growth. A zero value for the NFCI can be thought of as the U.S. financial system operating at a historically average level of risk, credit, and leverage (see chart 11).

Chart 10

Chart 11

While broader financial conditions' negative impulse on growth has eased since the banking sector turmoil in March--when three major U.S. banks failed and a global systemically important bank, Credit Suisse, failed for the first time since the global financial crisis--it remains significant historically (see chart 12).

Chart 12

Little Room For Cyclical Expansion

Broadly, the economy has little room for cyclical expansion. The output gap is positive (see chart 13), and the labor market is at full employment (see chart 14).

Chart 13

Chart 14

In August 2023, nonfarm payrolls outperformed expectations with a gain of 187,000 jobs, showcasing a resilient labor market, but the pace of growth has moderated (see chart 15). Furthermore, the employment-to-population ratio of prime-age workers (25-54)--a demographic unaffected by voluntary retirement--reached 80.9%, its highest since 2001 (see chart 16).

Chart 15

Chart 16

Meanwhile, industry capacity utilization has shown notable resilience in recent quarters, as evidenced by the marginal uptick to 79.7 in August 2023 (see chart 17), near the long-term average (1968-2022). Manufacturers' inventory stock is high relative to sales on a historical basis, while retailers are keeping it lean (see chart 18).

The total business inventory-to-sales ratio (for wholesalers) remained at 1.39 at the end of July 2023--a very high mark. A higher ratio implies a prolonged inventory turnover period, which may reflect reduced demand, often associated with a slowdown in economic activity.

Chart 17

Chart 18

Appendix

Definitions of positive/neutral/negative signals

• Positive: Overall economic activity will continue to expand in the near term, without an obvious slowdown.
• Neutral: Overall economic activity will continue to expand but may be slower. Right after a recession, a neutral signal indicates that the recovery may have just started.
• Negative: Overall economic activities will start to contract shortly afterward, roughly within a year.

The signals refer to the near-term growth perspectives suggested by the changes in a certain leading indicator during the past three months (since our previous publication). To determine whether the changes have been statistically significant (at a 5% significance level) to change the signal, we carry out a simple t-test against the historical sample. To capture the expansion and recession periods of business cycles without bias, we select the series between December 1982 and June 2009 as the historical representation, which includes three complete business cycles. For shorter series, we use the entire series until the end of 2017. Note that a negative growth signal is not equivalent to a recession in the near term.

Conventional interpretations

Although we use the traditional statistical methods, we are still aware of the widely used standards for interpreting the leading indicators.