It was designed as a real-time trigger for automatic fiscal support, not as a forecasting model, and its appeal is that it uses one series with almost no revision lag. Historically it has fired early in every US recession since the 1970s.
Its author has cautioned that the rule can misfire when unemployment rises because labour-force-participation and immigration are expanding labour supply rather than because demand is collapsing. Treat a trigger as a prompt to examine the composition, not as a verdict.
Example: the three-month average unemployment rate is 4.30% and the twelve-month low of that average was 3.70%. The gap is 0.60 points, above the 0.50 threshold, so the rule has triggered.
Related: unemployment-rate, recession, leading-indicator, labour-force-participation, nber-dating