NAIRU is unobservable and drifts with demographics, matching efficiency and the structure of the labour market. Estimates for the United States fell from around 6% in the 1990s to near 4.5% today, and those revisions were made after the fact.
The practical lesson from the late 2010s is humility: unemployment ran below every contemporaneous NAIRU estimate for years with no inflation acceleration, which suggested the estimate was simply too high. Treat it as a slow-moving anchor, not a trigger.
Example: unemployment is 3.8% against a NAIRU estimate of 4.4%, implying a tight labour market. With an Okun coefficient of 2, that 0.6 point gap maps to an output-gap near plus 1.2%.
Related: phillips-curve, unemployment-rate, output-gap, okuns-law, potential-gdp