Economists are confronting a paradoxical scenario where the labor market could enter a recession (job losses, rising unemployment) while the broader economy, measured by GDP, continues to expand. This potential disconnect challenges traditional definitions of an economic downturn and complicates forecasting.
The reliability of UI claims as a real-time barometer for job loss is diminishing. Stricter state eligibility rules post-pandemic, the prevalence of gig work as an alternative to filing, and high-wage tech layoffs where benefits are negligible all contribute to this indicator's declining usefulness.
Many economists who incorrectly predicted a recession in 2022-2023 now appear 'gun shy.' This recency bias may be causing them to avoid making a definitive recession call, even as negative economic indicators accumulate, leading to a reluctance to stick their necks out.
To formally change its baseline forecast to a recession, the firm employs a high-conviction rule of thumb. The internal probability must exceed two-thirds, ensuring there is a high degree of confidence and only a one-third chance of being wrong before making such a significant shift in outlook.
To navigate conflicting economic signals, Moody's built a model that uses a machine learning technique called a random forest. It aggregates 'votes' from numerous decision trees based on economic data, with labor markets carrying the most weight, to produce a single 12-month recession probability.
A historically reliable recession predictor, the Conference Board's Composite Leading Indicator, has been declining for years and experienced a peak-to-trough drop that has always preceded a recession. Its failure to correctly signal one in the 2022-2023 period shows how even trusted indicators can be fallible in the current economy.