Is a US recession more likely than you think?

Two recent events have inspired my latest blog post – the triggering of the Sahm rule and the release of the September Beige Book. As I update my latest Beige Book sentiment score, and using that in combination with assessing changes in the US unemployment rate, I’ve been trying to answer two questions:

1. What do the data readings imply for the probability of a US recession?
2. How does that compare with consensus?

Beige Book sentiment and unemployment jointly suggest a 70% chance of recession compared with Bloomberg consensus at 30%. I calibrate the model using NBER-dated recessions. The start of recession is not measured in real time but could be backdated by the Business Cycle Dating Committee if the economy subsequently moves into recession.

Early in the sample the recession probability series is noisy, which may be due to how the Beige Book was originally recorded. This typically occurred behind closed doors and the report was not publicly released until 1983.

Since around the end of 2022 the implied recession probability rose above 50% due to increased negative Beige Book sentiment. This aligned with consensus expectations. Yet currently, consensus expects a 30% chance of recession compared with an implied probability of 70%.

Recession forecasting is notoriously hard, and the baseline recession risk in any year is roughly one in six. Clearly the model does not attempt to capture idiosyncratic shocks to the economy, rather it focuses on classic economic slowdowns caused by the business cycle.

In terms of monetary policy, the market’s current pricing suggests rate cuts of 100bps through to year end. Whether recession materialises or the US Federal Reserve achieves a soft landing, we continue to monitor a range of key indicators.

But as a minimum, the latest Beige Book sentiment and unemployment trends seem to think that US recession risks are being materially underpriced by consensus.