The latest employment situation data came out today, and putting it in the context of various vintages of the FRBSF forecasts and the single (now two year old) vintage of the dynamic information equilibrium model (that was spot on) we can see that what was once FRBSF's pessimism in the future unemployment rate became an over-correction to optimism:
The most recent FRBSF forecast from January 2019 has already corrected against the optimism of last summer, revising their forecasts upward. It forecasts a basically constant unemployment rate at just under 4% through the end of 2022, but it's a bit late on the scene.
Instead of re-baselining my forecasts every couple of months, I didn't do an update of the forecast for almost two years and then did so only to compare against a new forecast from the CBO. This updated forecast for the past year  is getting the data right while the CBO, FOMC, and FRBSF are all looking a shade to optimistic for 2019:
 The big difference between the January 2017 forecast and the April 2018 forecast is that the newer forecast was able to better estimate the parameters (notably the size and duration) of the 2014 mini-boom.