It doesn't just do that. A careful reading of the establishment survey will tell you that the unemployment ratio is
unemployment = total jobs reported / people saying they're in the labour force (ie. who have a job or collect unemployment)
Part time jobs count as a fulltime job. So a poor person with 2 jobs is counted twice as employed, and long-term unemployed are excluded from the figures. It is hard to imagine, given the successive changes to this variable that this is an accident.
In reality the unemployment ratio should be:
people that are employed/people that could be employed
people that could be employed needs to be approximated, so I'm using the "working age population" figure.
I'm sure it's a total coincidence that the first ratio is 5.5%, showing a constant, if excruciatingly slow, decrease under the current administration (except during the GFC in 2008) and the second ratio is currently 25.8%, and shows a constant increase (except for the last 2 months) under the current administration. Caveat: I'm pretty sure the actual 25.8% figure is overstating matters, the changes in the figure are real: they represent either people who enter the workforce and can't find jobs, or people who get fired. Given anecdotes, and sentiment, I'm much more inclined to believe the latter : that there has been a constant increase of the unemployment figure since around 2006, extremely fast increase during 2008-2009, to ~12-13% followed by a slower increase after that, currently at 16% or so.
It is also funny that both figures start do diverge, immediately following a change in the way unemployment is calculated. One wonders ... which of those 2 things is cause, and which is effect.