Regarding direct job-to-job transitions, if you want to measure the number of job losses, these should presumably not be counted, since they are usually quits rather than job losses (although in some cases a laid off employee finds a new job before the old one ends, but I’m not sure there would be any way to identify those cases with any data series). In that respect the BED probably overcounts job losses, because in cases where the employee was not replaced at the old establishment, the transition would show up as a job loss at the old establishment and a job gain at the new establishment, even if it is actually a quit at the old establishment.

We have markets on all 50 states. The forecasts from each are summed up to produce an electoral college forecast.